[Senate Hearing 117-865]
[From the U.S. Government Publishing Office]
S. Hrg. 117-865
SECURING U.S. LEADERSHIP IN EMERGING
COMPUTE TECHNOLOGIES
=======================================================================
HEARING
BEFORE THE
COMMITTEE ON COMMERCE,
SCIENCE, AND TRANSPORTATION
UNITED STATES SENATE
ONE HUNDRED SEVENTEENTH CONGRESS
SECOND SESSION
__________
SEPTEMBER 29, 2022
__________
Printed for the use of the Committee on Commerce, Science, and
Transportation
[GRAPHIC NOT AVAILABLE IN TIFF FORMAT]
Available online: http://www.govinfo.gov
__________
U.S. GOVERNMENT PUBLISHING OFFICE
55-822 PDF WASHINGTON : 2024
-----------------------------------------------------------------------------------
SENATE COMMITTEE ON COMMERCE, SCIENCE, AND TRANSPORTATION
ONE HUNDRED SEVENTEENTH CONGRESS
SECOND SESSION
MARIA CANTWELL, Washington, Chair
AMY KLOBUCHAR, Minnesota ROGER WICKER, Mississippi, Ranking
RICHARD BLUMENTHAL, Connecticut JOHN THUNE, South Dakota
BRIAN SCHATZ, Hawaii ROY BLUNT, Missouri
EDWARD MARKEY, Massachusetts TED CRUZ, Texas
GARY PETERS, Michigan DEB FISCHER, Nebraska
TAMMY BALDWIN, Wisconsin JERRY MORAN, Kansas
TAMMY DUCKWORTH, Illinois DAN SULLIVAN, Alaska
JON TESTER, Montana MARSHA BLACKBURN, Tennessee
KYRSTEN SINEMA, Arizona TODD YOUNG, Indiana
JACKY ROSEN, Nevada MIKE LEE, Utah
BEN RAY LUJAN, New Mexico RON JOHNSON, Wisconsin
JOHN HICKENLOOPER, Colorado SHELLEY MOORE CAPITO, West
RAPHAEL WARNOCK, Georgia Virginia
RICK SCOTT, Florida
CYNTHIA LUMMIS, Wyoming
Lila Helms, Staff Director
Melissa Porter, Deputy Staff Director
George Greenwell, Policy Coordinator and Security Manager
John Keast, Republican Staff Director
Crystal Tully, Republican Deputy Staff Director
Steven Wall, General Counsel
C O N T E N T S
----------
Page
Hearing held on September 29, 2022............................... 1
Statement of Senator Cantwell.................................... 1
Statement of Senator Wicker...................................... 3
Statement of Senator Hickenlooper................................ 4
Statement of Senator Blackburn................................... 42
Statement of Senator Klobuchar................................... 43
Statement of Senator Lummis...................................... 45
Statement of Senator Baldwin..................................... 47
Statement of Senator Young....................................... 48
Statement of Senator Rosen....................................... 51
Statement of Senator Peters...................................... 54
Witnesses
Nancy Albritton, M.D., Ph.D., Frank and Julie Jungers Dean
College of Engineering Dean, University of Washington.......... 6
Prepared statement........................................... 8
Jack Clark, Co-founder, Anthropic, Co-chair, AI Index. Member,
National AI Advisory Committee................................. 10
Prepared statement........................................... 12
William B. (Trey) Breckenridge III, Director, High Performance
Computing Collaboratory, Mississippi State University.......... 18
Prepared statement........................................... 20
Steven C. Lupien, Director, University of Wyoming Center for
Blockchain and Digital Innovation.............................. 21
Prepared statement........................................... 23
Dr. Bob Sutor, Vice President, Corporate Development, ColdQuanta. 24
Prepared statement........................................... 26
Henry L. Jones II, Ph.D., Director, Research Development and
Scientific Entrepreneurship, University of Southern Mississippi 33
Prepared statement........................................... 34
Appendix
Response to written questions submitted to Dr. Nancy Albritton
by:
Hon. Maria Cantwell.......................................... 57
Hon. Kyrsten Sinema.......................................... 58
Hon. Raphael Warnock......................................... 60
Response to written questions submitted to William B. (Trey)
Breckenridge III by:
Hon. Maria Cantwell.......................................... 62
Hon. Kyrsten Sinema.......................................... 63
Hon. Raphael Warnock......................................... 64
Response to written questions submitted to Dr. Bob Sutor by:
Hon. Maria Cantwell.......................................... 65
Hon. Kyrsten Sinema.......................................... 67
Hon. Ray Ben Lujan........................................... 68
Hon. Raphael Warnock......................................... 69
Response to written question submitted to Henry L. Jones II,
Ph.D. by:
Hon. Maria Cantwell.......................................... 71
Hon. Kyrsten Sinema.............................................. 72
Hon. Raphael Warnock......................................... 73
SECURING U.S. LEADERSHIP IN EMERGING COMPUTE TECHNOLOGIES
----------
THURSDAY, SEPTEMBER 29, 2022
U.S. Senate
Committee on Commerce, Science, and Transportation,
Washington, DC.
The Committee met, pursuant to notice, at 10:04 a.m., in
room SR-253, Russell Senate Office Building, Hon. Maria
Cantwell, Chair of the Committee, presiding.
Present: Senators Cantwell [presiding], Klobuchar,
Blumenthal, Peters, Baldwin, Rosen, Lujan, Hickenlooper,
Wicker, Fischer, Blackburn, Young, and Lummis.
OPENING STATEMENT OF HON. MARIA CANTWELL,
U.S. SENATOR FROM WASHINGTON
The Chair. Good morning. The Senate Committee on Commerce,
Science, and Transportation will come to order. This morning,
we will be having a hearing on securing U.S. leadership in
emerging compute technologies, or in other words, what do we do
with the CHIPS and Science money and the actual appropriations
that we need to get to keep our competitive edge in compute
science.
So very distinguished panel here this morning to discuss
this very important issue. I would like to welcome all of them.
In August, the President signed the bipartisan CHIPS and
Science Act into law with historic commitment to U.S.
technology leadership.
This hearing is about ensuring what we do as a nation to
keep that commitment by building the workforce needed to stay
competitive in the most leading edge and consequential
computational disciplines.
Leading the world in computation grows the economy, creates
new jobs, and keeps America safe. Computing helped put
Americans on the Moon, develop faster and stealthier planes,
better weather forecasting artificial intelligence for
precision agriculture.
That is why the CHIPS and Science Act was so focused on
building America's computing capabilities. America knows that
the new law invested more than $50 billion into chip
manufacturing, and they already are seeing results.
The groundbreaking announcements in Ohio, Idaho, North
Carolina, and even in my home state of Washington have all put
manufacturing resurgence to a point of gaining steam. But just
as important in the law is the focus on research and workforce
development in ten key technology areas. Four of these areas,
artificial intelligence, semiconductors, quantum science, and
distributed ledger technologies deal with improving
computation.
Job openings in these areas are soaring, but the number of
workers definitely is not keeping pace. It is a crisis. That is
why the CHIPS and Science Act authorized $13 billion for STEM
education, including funds for nearly 40,000 scholarships,
fellowships, and traineeships. That is why it has money for
faculty hiring and training.
And make no mistake, America's workforce shortages are
serious. And failing to make these investments and failing to
retain the talent from around the world is not an option. With
six decades of AI development, computers now operate vehicles,
translate languages, create art, help on the factory floor, and
most of you know in this room are probably hearing AI
technologies right now.
But AI doesn't work alone, and it is far from perfect. It
must be trained by humans and must often be partnered with the
human in doing the work, and that means we need a generation of
AI literate workers. A recent Georgetown study supported that
we will add 1 million AI jobs between 2019 and 2029.
AI requires vast amounts of information and excellent data
retrieval capabilities, which is one of the promises of
distributed ledger technologies. But in the computation--but in
the computer science foundational discipline for this
technology, universities are turning away students because the
supply of teachers isn't keeping up with demand.
The country that combines the power of quantum computing
with artificial intelligence could make an insurmountable leap
forward in technology and rewrite the rules of the road for
cybersecurity, for creating medical innovations and saving
thousands of lives, and developing battlefield technologies.
The nation faces a shortage of a quantum talent, with fewer
than 5 percent of U.S. PhDs in relevant fields focusing on
quantum science. So the stakes are high. International
competition is mounting. Funding for CHIPS and Science must not
stop with the appropriations for chip manufacturing.
America needs access to better chips, but it also needs the
research and workforce to put those chips to use. So that is
why I am proud to have with us one of our witnesses, Dr.
Albritton, who we will hear from shortly. Washington is a
leader in emerging computational technology.
Seattle has the Nation's third largest AI workforce. The
University of Washington is one of the Nation's top
institutions for AI. In the Spokane region, home to more than
7,000 farms and ranches, WSU leads the Institute on Agriculture
AI with a focus on AI enabled workforce, and companies like
Amazon, Microsoft, T-Mobile and Starbucks are applying
distributed ledger technologies to telecommunications and
supply chain.
And we are building the quantum workforce through the
efforts like the University of Washington's Quantum X
Institute, which is creating a graduate certificate program in
quantum science. With the CHIPS and Science Act, I hope to see
Washington continue to grow this leadership and developing a
workforce.
Each of these technologies will face challenges, but if
America is to lead, we need to continue to lead in talent. So I
look forward to hearing from the distinguished panel today, and
I will now turn to the Ranking Member for his opening
statement.
STATEMENT OF HON. ROGER WICKER,
U.S. SENATOR FROM MISSISSIPPI
Senator Wicker. Thank you, Madam Chair. And let me say of
all the legislation that you and I have worked shoulder to
shoulder on, I can't think of anything more significant for the
future of our country, and not only our economy but also our
national security, than the CHIPS and Science Act.
And so thank you so much for being my partner in that
regard and for persuading the House and Senate to include the
very significant additions that we made. Today's hearing on
securing U.S. leadership in emerging technologies could not be
more timely.
The recently enacted CHIPS and Science Act provided
critical investments and policy tools to advance American
innovation in key technologies, including, as the chair said,
quantum computing, artificial intelligence, and blockchain.
The development of these emerging technologies is important
but also poses important questions for policymakers involving
security, privacy, and other impacts on society. Today's
witnesses play key roles in technology development and
innovation.
The Chair is proud of witnesses from the state of
Washington, and I am similarly delighted that Dr. Henry Jones
is here, the Director of Research and Development and
Scientific Entrepreneurship at the University of Southern
Mississippi, and Mr. Trey Breckenridge, the Director of High
Performance Computing at Mississippi State University.
I look forward to hearing their thoughts on spreading the
geography of innovation when it comes to innovation. The--one
of the most significant provisions that we were able to add,
Madam Chair, is to spread the geography and avoid overreliance
on a handful of States and big universities. American
leadership requires that we take advantage of the talent,
expertise, and capabilities found throughout America.
In that vein, I would appreciate eyewitnesses? perspectives
on where the United States ranks regarding the development of
these emerging technologies, particularly as compared to China,
and what more should be done to secure a preeminent position.
China and other nations are increasingly dominant in technology
innovation, posing a massive threat not only to our economy,
but, as I said, to national security.
There is no more important competition than the one for
technological supremacy between the United States and China.
Congress has recognized this fact and has taken the first step
by passing the CHIPS and Science Act, which would position the
United States to be a global leader once again.
This law establishes a new Directorate for Technology,
Innovation and Partnerships at the NSF to accelerate the
process of translating basic research into technology
development for commercial use and making America more
competitive globally. Geographic diversity is a vital--is vital
to American innovation and enlisting the talent and expertise
of STEM researchers nationwide is an important priority.
In the CHIPS and Science Act, the law guarantees that
EPSCoR, a program designed to stimulate competitive research in
25 predominantly rural states, will receive 20 percent of all
R&D funding from the NSF, up from the current 13 percent. This
will go a long way toward developing--toward leveraging the
whole of America's technological expertise in this global
contest.
Make no mistake, ensuring that America leads the world in
tech is good for our economy and national security, and it is
also good for the cause of freedom, stability, openness, and
our democratic values around the world. With the right vision
and priorities, we in Congress can make sure the 21st century
is another American century. So I look forward to hearing from
our distinguished witnesses, and I thank the Chair.
The Chair. Thank you, Senator Wicker. And thank you for
your leadership on the EPSCoR issue, particularly. It was
something that we spent a lot of time on. And I firmly believe
I saw this morning that Steve Case has a book, ``The Rise of
the Rest.''
And I think with the Intel investment in Ohio and what we
are doing, I don't want to steal Dr. Albritton's testimony, but
I think she is going to tell us she is originally from
Louisiana, and I think that is the point, that we have talent
all throughout the United States of America and what do we do
to better unleash that.
And hopefully today's hearing will be a very poignant point
forward to our appropriators that we have to get the rest of
this right. But now I am going to----
Senator Wicker. Madam Chair, if because of your statement,
sales of Mr. Case's book shoot through the roof, the Case
family will be able to have Christmas this year.
The Chair. I am pretty sure they can anyway, but I probably
would have had him come in--this is a hard subject because it
is not hard, it is distributed generation. But when we were
definitely in the negotiations, we spent a lot of time on this
for a lot of different reasons. And some institutions who are
already making R&D contributions, it is hard.
But I think we got it right and I think that we are going
to move forward, and I think we are going to see the dividend
of that. So thank you for your leadership. Senator Hickenlooper
from the Subcommittee is going to make an opening statement.
STATEMENT OF HON. JOHN HICKENLOOPER,
U.S. SENATOR FROM COLORADO
Senator Hickenlooper. Thank you, Madam Chair. I thank all
of you for schlepping out here and spending the better part of
a day with us. It is exciting to be part of such an impressive
group. The recently passed CHIPS and Science Act really is a
massive investment in our country's future.
I was proud to be a conferee on the bipartisan Innovation
and Competition Conference, part of that. The drive that R&D
receives is what helps our economy grow, it improves our public
health, it strengthens our national defense. It affects us in
almost every manner possibly.
When I was a kid, there was a movie called, Being There,
written by a guy named Jerzy Kosinski who, Chance the gardener,
turned out to be an expert in the economy. And basically he was
rather simple minded. But his metaphor he used for everything
was back to the garden, whether it had the nutrients.
The garden as an ecosystem. Well, I think what the CHIPS
bill does is invest $200 billion into our scientific ecosystem.
And just like the garden in Being There, the nutrients and the
diversity of what is in that ecosystem is crucial to its
success. We need to grow and diversify our skilled workforce,
making sure we have pathways other than the traditional 4-year
degree.
We have to increase manufacturing and innovation in every
part of the nation, as Madam Chair was just saying. Steve Case
and others have begun that process of looking into rural areas
all across America and finding out how do we stimulate
innovation everywhere. We are in a fierce rivalry, a fierce
competition with China, around a number of countries, and we
need everybody on deck.
We are thrilled to see last week that the Senate confirmed
Dr. Prabhakar to head the White House Office of Science and
Technology Policy. She is going to play a pivotal role in
implementing CHIPS and Science by developing the National
Science and Technology Strategy, and perhaps more importantly,
coordinating the efforts across all the Federal agencies and
with the different offices of the White House.
I think what we are doing here today is really discussing
the future of computing to a large extent. We have three
focuses today that showcase what we can already achieve, but
also what we will be able to do in the near future. One,
artificial intelligence, AI, has the power to automate complex
processes and improve our efficiency in every sector of our
economy.
As more and more businesses incorporate AI into their
operations, we need to accelerate that integration. We must
ensure that these systems are transparent, and as much as
possible, free from bias.
I applaud the National AI Initiative Office's efforts to
gather broad public input and to convene experts from
Government, academia, and industry to make sure we get off on
the right path. Second, quantum information science, UIS, is an
advanced field, which I am proud to say, deep history in
Colorado. It will transform the future of computing and let us
run simulations with staggering speed.
Colorado's own ColdQuanta, Dr. Sutor is here, is on the
cutting edge, some would say the bleeding edge, of this
computing revolution. NIST has a partnership with CU Boulder to
advance quantum technologies through their joint institute, the
JILA, including world record accurate atomic clocks.
Further partnership with CU Boulder, NIST, and front range
companies has joined together to form the CUBIT, I did not come
up with that acronym, CUBIT initiative, which is going to help
accelerate our growing quantum ecosystem in Colorado, and I
think in a similar way, we will accelerate ecosystems around
the country.
Third, after quantum information science's distributed
ledger technology, DLT, often implemented as blockchains, is
more popularly known as blockchains, can improve security of
financial transactions, support innovations and data privacy.
All kinds of implementations, health care just being one.
A major DLT innovation hub is located in Wyoming, the home
of space and science subcommittee's Ranking Member, Senator
Lummis. Today, I look forward to discussing important topics
for all these technologies. They are so important to Colorado
and to the Nation.
Which barriers should we break down to make sure that we do
grow a skilled STEM workforce in all these fields? What are the
guardrails we need to make sure we create to ensure these
technologies are developed transparently, used ethically, and
benefiting everyday lives of all Americans.
With the springboard created by the historic CHIPS and
Science Act, how do we ensure that the United States remains
the global gold standard for innovation in each of these
fields? So I thank all of you for joining us in delving into
these issues and I return--yield to the Chair.
The Chair. Thank you. I don't see Senator Lummis here or
online, but if she does arrive and wants to give us opening
statement as the Ranking Member of the Subcommittee, we will
definitely look forward to hearing her comments. We will now
turn to the witnesses. I want to welcome Dr. Nancy Albritton,
Dean of the College of Engineering at the University of
Washington, the Frank and Julie Jungers Dean.
I also want to welcome Dr. Jack Clark, Co-Founder of
Anthropic; Mr. William Breckenridge III, Director of High
Performance Computing and High Performance Computing
Collaborative, Mississippi State University; Mr. Steve Lupien
is joining us remotely. He is the Director of the Center for
Blockchain and Digital Initiatives at the University of Wyoming
College of Business.
Dr. Bob--is at Suter? Sutor--Dr. Bob Sutor, Vice President,
Corporate Development of ColdQuanta; and Dr. Henry Jones,
Director of Research and Development in Scientific
Entrepreneurship, University of Southern Mississippi. So
welcome to all of you. We will start with you, Dean Albritton.
STATEMENT OF NANCY ALBRITTON, M.D., Ph.D., FRANK AND JULIE
JUNGERS DEAN COLLEGE OF ENGINEERING DEAN, UNIVERSITY OF
WASHINGTON
Dr. Albritton. Good morning, Chairman Cantwell and Ranking
Member Wicker, and distinguished members of the Committee.
Thank you, Senator Cantwell, my Senator, for the opportunity to
testify about the value and continued importance of our Nation
investing in emerging computing technologies.
As noted, I am the Dean of the University of Washington
College of Engineering, and the three technologies the
Committee is focusing on today are areas where the University
of Washington is making significant contributions. Indeed, our
faculty lead multiple large, multi-institutional national
science foundation awards to advance the potential of AI,
machine learning, and data science, all while training the next
generation of workers and innovators.
As we have already noted, Congress took a bold step with
the CHIPS and Science Act. I am encouraged by this step. A
sustained Federal investment in these programs is essential for
our Nation to maintain its leadership on this fierce global
landscape. But it is also important to leverage collaborative
opportunities between Government, academia, and industry, and
to build a workforce that reflects the rich diversity of our
Nation.
This morning, I will focus my comments on the impact of
investment in quantum information science. Starting with an
example and its promise. As we all know, Hurricane Ian has made
landfall in our Southern States, and as noted, I am a native of
Louisiana. I keenly understand the toll that hurricanes have on
a community and also the hardship of rebuilding.
Thankfully, though, forecasting models developed by NOAA,
NASA, and the Pacific Northwest National Laboratory help us
predict the path and intensity of these hurricanes. It is a
tremendous improvement from my childhood, but just imagine if
we could prepare for storms with a much higher degree of
certainty.
Theoretically, a full scale quantum computer can weather--
perform weather forecasting with high accuracy by handling vast
amounts of data in seconds, saving lives and reducing property
damage. In the 1980s, classical computing and personal
computers changed the world. But now, advances in quantum
information science promise major breakthroughs in
communications, computing, and simulation.
Just as quantum science has enabled groundbreaking
technologies such as GPS, MRI, lasers for health care
applications, the realization of quantum information science
will fundamentally change the way that we live and work. But
now is the time to accelerate Federal investment to maintain a
competitive edge.
Across the globe, growing numbers of universities,
including the University of Washington, have established major
quantum information programs, and the competition for students,
researchers, faculty, and funding is intense, as is industry
hiring. The new quantum information initiatives in the U.S. and
around the world all have the same goal, not to miss this
window to lead in this area.
But no country is better positioned to emerge in the top
cohort than the U.S. is with its partnerships between academia,
Government, and industry. At the University of Washington,
though, the lack of critical quantum information science
capabilities, particularly those associated with the large cost
of implementation and maintenance, hampers our impact in this
growing field.
Another challenge to industry growth is the shortage of
equipped and diverse workforce. It is already difficult to fill
skilled STEM job openings, and as quantum computing investments
grow, competition for skilled worker intensifies and is an
unsustainable demand.
At the University of Washington alone, student demand far
exceeds capacity and most of our STEM programs, forcing us to
turn away excellent students when our Nation demands a skilled
workforce.
America's research universities stand ready to partner with
you to provide leadership, research, and workforce education. I
ask that you continue to sustain and increase Federal funding,
particularly in those science agencies that will enable us to
remain a global quantum information leader.
But more specifically, increase Federal investment in
workforce development and education, and accessible quantum
testbeds and quantum cloud computing, and high risk engineering
and science research and more fundamental quantum information
research, but also an investment in technology policy will be
particularly impactful.
Thank you for your time and consideration. I welcome your
questions.
[The prepared statement of Dr. Albritton follows:]
Prepared Statement of Nancy Albritton, M.D., Ph.D., Frank and Julie
Jungers Dean College of Engineering Dean, University of Washington
Good morning, Chairman Cantwell and Ranking Member Wicker, and
distinguished Members of the Committee. Thank you, Senator Cantwell, my
Senator, for the opportunity to testify about the value and continued
importance of our Nation investing in emerging computing technologies.
I am the Dean of the University of Washington (UW) College of
Engineering, and the three technologies the committee is focusing on
today--artificial intelligence (AI), quantum information science (QIS),
and distributed ledger technologies (DLT)--are areas where the
University of Washington is making significant contributions and
leading nationally and internationally to develop and discover the
potential of these fields while training the next generation of workers
and innovators.
Earlier this year Congress took a bold step to ensure that the
United States is equipped to be a global leader in science and
innovation with the passing of the CHIPS and Science Act. As you know,
this legislation is designed to revitalize American science and
innovation, build a strong and diverse STEM workforce, create solutions
for the climate crisis, and support American manufacturing. As a leader
in academia, I am encouraged by this critical Federal funding, and I am
here today to urge you to continue to invest in our Federal science
agencies and initiatives empowered by CHIPS. Sustained Federal
investment in these programs are essential for our Nation to remain a
leader in a fierce global landscape, to leverage opportunities for
collaboration between government, academic, and business sectors, and
to build a workforce that reflects the rich diversity of our Nation.
The University of Washington has significant expertise in the three
areas that the committee is focusing on today. UW faculty serve as PIs
on multiple large, multi-institution NSF awards in artificial
intelligence, machine learning, and data science, including the
Institute for Foundations in Data Science, the Institute for
Foundations in Machine Learning, and the AI Institute for Dynamic
Systems. However, this morning I will focus my comments on the impact
of investing in quantum information science and I'll start with an
example of its promise. As I write, Hurricane Ian is strengthening
before making landfall in our southern states and coastlines, with the
potential to shatter communities. As a native of Louisiana, I keenly
understand the toll that a destructive hurricane can have on a
community and the hardship of rebuilding. Thankfully, forecasting
models help us predict the path and intensity of hurricanes. Large
supercomputers and artificial intelligence already aid forecasting,
such as models developed by NOAA, NASA and the Pacific Northwest
National Laboratory (PNNL) that can predict when hurricanes will
rapidly intensify. It is a tremendous improvement from when I was
growing up, but we can and should do more. However, to achieve greater
accuracy in weather forecasting, more computational power is needed.
Theoretically, a full-scale quantum computer can improve weather
forecasting methods by handling huge amounts of data (terabytes per
second) containing many variables and optimizing complex algorithms,
and can do so in seconds. With quantum computing, we will be able to
prepare for these storms with a much higher level of certainty,
potentially saving lives and reducing property damage.
Our society stands on the brink of a major revolution driven by
quantum information science with boundless potential to fundamentally
change the way we live and work. Quantum information science uses
quantum effects to acquire, transmit, and process information. Quantum
science has already enabled us to better understand nature and advance
groundbreaking technologies like GPS, MRI scans, and lasers for
healthcare applications such as eye surgery and joint surgery.
Like the 1980s when classical computing and the personal computer
changed the world, recent advances in quantum information science
promise major breakthroughs in communications, computing, and
simulation. Industry and governments around the world are investing
heavily in quantum information science, recognizing its potential and
poising to capitalize on it. As the U.S. aims to be a scientific leader
of the coming quantum information age, now is the time to accelerate
Federal investment, as outlined by CHIPS, so our Nation is a global
leader in this field.
Currently, there are growing numbers of universities with
established major quantum information programs in the world, and the
University of Washington is proud to be one such program with robust
partnerships with industry including Microsoft, Boeing, Google, and
Amazon, and with the Pacific Northwest National Laboratory (PNNL).
Internationally, quantum information programs are underway in Sweden
(WACQT), the Netherlands (Quantum Delta), Japan (Moonshot), Israel
(Israel National Quantum Initiative), and the U.K. (UKNQTP). Germany,
France, Austria, and Canada are also substantially investing in this
area. China is making massive investments in quantum computing, and
quantum technology more broadly, making it certain that they will
emerge as a leader in this area in the next decade. Worldwide, the
competition for top students, researchers, faculty, and funding is
fierce, as is hiring by companies. The new quantum information
initiatives in the U.S. and around the world have the same goal: not to
miss the window to lead in this area. No country is better positioned
to emerge in the top cohort than the U.S. in partnership with academia,
government, and industry.
Investment in America's leading research universities will allow
talented faculty and students to further innovative science that will
elevate the U.S. as a global destination for knowledge and discovery in
quantum information sciences. These foundational investments will
influence economic and national security, prepare U.S. students for
jobs with quantum information technology, enhance STEM education at all
levels, and accelerate exploration of quantum information frontiers,
all while expanding and diversifying the talent pool for the industries
of the future in Washington state and across the Nation.
For example, the Washington Quantum Technologies, Teaching and
Testbed Laboratory (QT3) provides a regional resource for hands-on
quantum technology training to the next generation of quantum
scientists and engineers and state-of-the art quantum device
characterization research tools in a publicly accessible user facility.
In this second quantum revolution, society will leverage the
quantum-mechanical properties of light and matter to enable new
technologies in computation, communication, and sensing. Federal
funding to enable universities to train the next generation of
scientists and engineers is needed to enable this revolution. To
realize practical quantum technologies, quantum expertise is needed
through the full quantum stack--from materials, devices and hardware to
software and algorithms. DOE NQI centers have been established to drive
forward the research and training in select laboratories. These
centers, however, do not address the need for shared quantum
infrastructure for training and testbeds where Federal funding would
serve a critical need.
At the University of Washington, the lack of critical capabilities,
primarily originating from the large cost of implementing and
maintaining them, hampers our impact. While we have targeted systems
that work at room temperature, the leading quantum computing platforms
work at ultra-low temperatures, just a fraction of a degree above
absolute zero. Our researchers and students need access to this
capability. We are able to provide our current capabilities by focusing
on a single qubit platform which can function at room temperature at
one particular energy. This particular platform at room temperature
cannot scale to the many qubits required for meaningful quantum
computation. We also need to seek to expand to enable the excitation
and detection capabilities to discover new qubits which have the
potential to scale even further.
The Boston Consulting Group estimates that quantum computing alone
could create a value of $450 billion to $850 billion in the next 15 to
30 years if the technology scales as fast as predicted \1\. Quantum
information science presents a tremendous economic opportunity and a
substantial hurdle. One of the biggest issues impeding the growth of
this industry is the shortage of an equipped quantum information
science workforce. This shortage has a significant impact on the future
growth of industry. In Washington state workforce development will
impact the recruitment to existing Washington companies who have
expanding footprints in quantum computing including Microsoft, Boeing,
Amazon and Google as well as to new start-ups. It is increasingly
difficult to fill skilled STEM job openings, which is further
compounded for companies where advanced degrees in physics, chemistry,
materials science, engineering and computer science are needed. As the
investment and interest in quantum computing grows, competition for
skilled workers is intensifying and creating an unsustainable demand.
This demand cannot be addressed without an accelerated and unrelenting
investment in training and development.
---------------------------------------------------------------------------
\1\ Jean-Francois Bobier, Matt Langione, Edward Tao, and Antoine
Gourevitch, ``What Happens When `If' Turns to When in Quantum
Computing,'' Boston Consulting Group, 2021, https://web-assets.bcg.com/
89/00/d2d074424a6ca820b1238e24ccc0/bcg-what-happens-when-if-turns-to-
when
-in-quantum-computing-jul-2021-r.pdf.
---------------------------------------------------------------------------
According to a recent report from the Washington Roundtable, a
nonprofit organization comprised of executives of major private sector
employers in Washington state, in the next five years ``Washington
state's anticipated annual job growth rate of 2.3 percent will far
outpace the national rate of 1.3 percent. Seventy percent of these jobs
will require a post-high school credential. Washington employers want
to hire local talent to fill these positions whenever possible and it's
essential that our young people are ready.'' \2\ At the University of
Washington alone, student demand far exceeds capacity in the UW's
engineering and natural sciences (including computer science and
engineering, mathematical, physical and life sciences) programs. We are
forced to turn away excellent students while our Nation demands a
skilled workforce.
---------------------------------------------------------------------------
\2\ Washington Roundtable, January 2022, https://
www.waroundtable.com/wp-content/uploads
/2022/02/WRT_PostsecondaryEnrollmentCrisis_Report_1.2022.pdf.
---------------------------------------------------------------------------
Failing to prepare our citizens for the innovation economy
compromises our Nation's long-term competitiveness and economic
stability and disadvantages our citizens and communities. Industry,
government and universities must step up and invest in the quantum
workforce. As we enter an increasingly specialized economy, America's
leading research universities, including the University of Washington,
are uniquely poised to provide leadership, research, and workforce
education to meet this need, but we need Federal investment.
To be competitive we need to educate the workforce of the future
and we know diverse teams lead to better results, so we are investing
in programs to expand access to more Washington students, as well as
asking for investment from the state. Students from low-income
backgrounds, underserved public high schools, as well as first-
generation college students, are particularly likely to suffer from
financial, social, and emotional challenges, and struggle to adjust to
college life and expectations. Our goal is to grow the infrastructure
needed to support these students, many of whom arrive with little
background or outside support necessary to navigate the rigorous
coursework required for engineering and computer science. We measure
our success in the students who graduate and move on to thriving
careers in our Nation's industries. Central to our public mission, we
strive to identify these students, recruit them, and enable them to
succeed.
STEM creates a future of opportunity. It's a pipeline of local
talent that will serve students, businesses, and communities across the
nation, with benefits for decades. Through key relationships with
industry, government, and academic partners, our Nation's universities
can connect the best and brightest minds to advance quantum technology
faster. The success of quantum information is closely aligned to
ongoing fundamental science, which is why Federal investment is
required. Federal leadership and investment are the foundation for
these advances for all. As a representative of academia, we stand ready
to partner with you. I ask that you continue to accelerate discovery
through sustained and increased funding of the Federal agencies that
enable us to remain a quantum information leader in a fierce global
landscape. And I leave you with five areas where Federal investment
would be particularly impactful: Increased funding for workforce
development and education, support to develop accessible quantum
testbeds and quantum cloud computing for all, increased funding for
high-risk engineering and science research given the remaining
technological barriers for quantum information science to become
broadly usable, and more fundamental quantum information research and
investment in technology policy.
Thank you for your time and consideration.
The Chair. Thank you, Dr. Albritton. And we look forward to
asking you questions on that. And now we welcome Mr. Clark.
STATEMENT OF JACK CLARK, CO-FOUNDER, ANTHROPIC,
CO-CHAIR, AI INDEX, MEMBER, NATIONAL AI ADVISORY COMMITTEE
Mr. Clark. Chair Cantwell, Ranking Member Wicker, and
members of the Committee, thank you for inviting me to testify
today, and thank you for passing the CHIPS and Science Act.
Your work is helping to set the United States up for continued
leadership in the development of transformative technologies
such as artificial intelligence.
For this testimony, I will make recommendations to help the
U.S. meet the challenge of international competition, ensure
better collaboration between industry, Government, and
academia, and grow a diverse workforce to meet the needs of our
economy, all while ensuring the safe and responsible adoption
of AI.
My recommendations are first, the U.S. should fully deliver
the CHIPS and Science Act and make further investments in the
measurement and monitoring of AI, both here and abroad. We need
to know if we are in the lead, and if not, who is.
Second, the U.S. should build experimental infrastructure
for the development and testing of AI systems by academic and
Government's users, and so should be ambitious in how it
supports the proposed National AI Research Resource. And third,
we must prioritize the creation of testbeds for AI across the
country, which can be used to train a new, diverse workforce in
the art of assessing and deploying AI systems.
The U.S. today enjoys an enviable position in AI. We have a
strong academic base, thriving commercial sector, and this
hearing is an example, a Government that cares about how to
support the industry.
AI systems have already made it out of the lab and are
being used in the world. GitHub, a subsidiary of Microsoft,
recently developed a tool called ``Copilot,'' a spell checker
for code. It helps programmers program better, and those that
use Copilot are 50 percent faster than those that don't.
Researchers at Stanford have combined satellite imagery and
machine learning to measure sustainable development outcomes in
areas such as hunger relief, population density, and economic
activity. And meanwhile, Google researchers have built
translation technology for languages for which there is scarce
data.
And using this, they have added 24 new languages to their
translate service, letting all of us talk more to each other.
And further afield, a new class of AI systems called Foundation
Models are giving us the AI equivalent of Swiss Army knives,
single software applications that can perform a multitude of
tasks ranging from generating text and code, to helping people
to create images and songs, to summarizing vast documents such
as legislation.
But the U.S. can't rest on its laurels. AI is a competitive
technology, and China already rivals for U.S. in AI R&D. In
2021, China published more AI research papers from the United
States and filed more patents than any other country. The U.S.
still holds the lead in the number of accepted papers and also
in the number of citations its papers get but the gap is
closing.
China has also proved capable of significant feats of AI
engineering and research. Chinese scientists have won image
recognition challenges, which were previously dominated by
American companies. And Chinese companies like Huawei have been
the first entities to publicly replicate frontier research
published by American companies. So what do we do? I have two
specific proposals.
As someone who spends their days in an AI lab, I can tell
you that testing and evaluating AI systems is fundamental to
realizing that commercial applications and identifying any
safety issues.
Therefore, we must ensure that the National Institute of
Standards and Technology is able to stand up AI testbeds across
America so local communities can take AI systems out of the lab
and vigorously test and deploy them. Second, we need better
experimental infrastructure, here, big computers, because that
is fundamental for research. For that reason, I wholeheartedly
support the National AI Research Resource.
If we want America to benefit from AI, then we need to make
it easier for our Nation's best scientists to build and
experiment on it. And that is an opportunity the NAIRR gives
us. In closing, the U.S. remains in a strong leadership
position in the AI ecosystem, but China continues to close the
gap.
For the U.S. to maintain its leadership, it should ensure
adequate funding for AI research, computational infrastructure,
and support for the measurement and assessment of AI systems.
Ensuring strong domestic AI capabilities is paramount to
protect our national security interests and ensure our
continued economic competitiveness.
And Senators, I used an AI to write the last few sentences
of this testimony. Thank you.
[The prepared statement of Mr. Clark follows:]
Prepared Statement of Jack Clark, Co-founder, Anthropic, Co-chair, AI
Index, Member, National AI Advisory Committee
How testing and experimental infrastructure will let the United States
take advantage of the industrialization of AI.
Chair Cantwell, Ranking Member Wicker, and members of the
committee, thank you for the opportunity to speak with you today about
the important topic of how the United States can maintain its
leadership in emerging compute technologies. First, thank you for
passing the CHIPs and Science Act. Through passing this, you have
helped to set the United States of America up for continued leadership
in the development and deployment of transformative technologies such
as artificial intelligence.
For this testimony, I will make a few simple recommendations, which
I hope will help us meet the challenge of international competition;
increase opportunities for collaboration between the government,
industry, and business sectors; and build a diverse and inclusive
workforce to meet the growing demands of our evolving economy--all
while ensuring the safe and responsible adoption of technology.
These recommendations are as follows:
The United States should fully fund the CHIPS and Science
Act, and make further investments in the measurement and
monitoring of the artificial intelligence development ecosystem
both domestically and abroad. Having accurate information about
progress within the United States, among our allies, as well as
progress occurring in other countries, is crucial for making
good decisions about American technology strategy. We need to
know if we're in the lead or if we're coming from behind, and
where any gaps may be.
The United States should seek to develop experimental
infrastructure at scale for the development and testing of
artificial intelligence systems by academic and government
users. Concretely, the proposed National AI Research Resource
can be best leveraged by pairing it with the creation of
testbeds which can be used to train a new, diverse workforce in
the important work of developing and assessing AI systems for
economic applications and safety assurance.
The United States should prioritize the development of
resources for the assurance of AI--specifically, tools for the
testing, evaluation, and benchmarking of artificial
intelligence systems. The better we get at AI assurance, the
more confidence we can have in AI systems, and the more we can
create opportunities for collaboration across the private
sector, government, and academia. Additionally, as more
communities have the ability to test out different
applications, they'll develop new products and services along
the way.
Before I expand on these recommendations, I'd like to state why
they're necessary, and why I'm so appreciative you are having a hearing
about this now.
HOW WE GOT HERE
First, I want to provide an update on just how rapidly the field
has been advancing. The past decade has been distinguished by what we
can think of as the industrialization of artificial intelligence; AI
has gone from an interesting topic of research and discussion among
researchers, to something of real economic and strategic utility.
We can very roughly draw the ``ignition point'' for the
industrialization of AI to 2012: this is when a team of researchers at
the University of Toronto were able to win a highly-competitive image
recognition competition known as ImageNet using a then-novel
technique--taking a bunch of so-called neural networks, layering them
on top of one another like a lasagne, and then training them on a
significant amount of data \1\. The result was a system which set a new
state-of-the-art on image recognition and which led to significant
investments in AI by industrial actors here and abroad.
---------------------------------------------------------------------------
\1\ ImageNet Classification with Deep Convolutional Neural
Networks, https://papers.nips.cc/paper/2012/hash/
c399862d3b9d6b76c8436e924a68c45b-Abstract.html, 2012. One of the team
members, Ilya Sutskever, went on to help found OpenAI, a major AI
research company based in the United States.
---------------------------------------------------------------------------
Since then, the field has notched up a few notable achievements.
Systems like ``AlphaFold'' \2\ have revolutionized the field of protein
structure prediction, which is a key input to scientific development.
Other AI systems have proven better able to stabilize the plasma in
fusion reactors than any human or previous software system \3\.
Meanwhile, AI has begun to make its way to the consumer so quickly that
many do not realize they're already interacting with it throughout the
day: voice recognition systems have improved substantially, we're all
able to search through the photos on our phones now to find pictures of
our dogs, cats, and family members.
---------------------------------------------------------------------------
\2\ AlphaFold: a solution to a 50-year-old grand challenge in
biology, https://www.deepmind
.com/blog/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-
biology, 2020.
\3\ Accelerating fusion science through learned plasma control,
https://www.deepmind.com/blog/accelerating-fusion-science-through-
learned-plasma-control, 2022.
---------------------------------------------------------------------------
These developments are fantastically exciting. Remember, ten years
ago, none of these things were possible. Now they are. AI is now being
applied in a vast range of fields, and we've barely scratched the
service. Just to give you an idea of what is possible, here are a few
examples of how AI is being applied today and the positive impact it is
having on the world:
Enhancing developer productivity: GitHub's Copilot product--
an auto-completion tool used for computer programming tasks--
has been shown to make software developers more than 50 percent
faster in their work than developers that do not use the tool.
The study also found higher levels of developer satisfaction,
with 60-75 percent of users feeling less frustrated by daily
programming tasks \4\ (GitHub).
---------------------------------------------------------------------------
\4\ Quantifying GitHub Copilot's impact on developer productivity
and happiness, https://github.blog/2022-09-07-research-quantifying-
github-copilots-impact-on-developer-productivity-and-happiness/, 2022.
Estimating sustainable development outcomes: Researchers at
Stanford University have demonstrated how combining satellite
imagery and machine learning can help measure sustainable
development outcomes in areas such as hunger relief, population
density, and economic activity \5\ (Stanford, Science)
---------------------------------------------------------------------------
\5\ Using satellite imagery to understand and promote sustainable
development, https://www
.science.org/doi/10.1126/science.abe8628?cookieSet=1, 2021.
Measuring agricultural health: Academic researchers have
developed an image recognition system for detecting
agricultural diseases in the cassava plant (a critical food
source for millions of people across Africa), using just a
mobile device \6\ (arXiv).
---------------------------------------------------------------------------
\6\ See: Using Transfer Learning for Image-Based Cassava Disease
Detection, https://arxiv
.org/abs/1707.03717, 2017.
Low-resource language translation: Google researchers have
found a way to develop translation technology for
underrepresented languages using only text in the original
language (traditional machine translation systems typically
work with two sets of text: the original language text, and its
translation to the target language). Using this new approach,
Google was able to add 24 under-resourced languages to its
Translate service and develop a repeatable method to include
other languages from around the globe \7\ (Google Research,
arXiv).
---------------------------------------------------------------------------
\7\ See: Building Machine Translation Systems for the Next Thousand
Languages, https://arxiv.org/abs/2205.03983, 2022.
There are also exciting developments at the frontier; in the past
few years, so-called ``Foundation Models'' \8\ have emerged which show
how AI is moving from an era of dedicated and specific tools to models
that behave more like ``Swiss Army knives''--a single model will be
able to do a broad range of tasks, many of which are scientifically and
economically useful.
---------------------------------------------------------------------------
\8\ On the Opportunities and Risks of Foundation Models, https://
fsi.stanford.edu/publication/opportunities-and-risks-foundation-models,
2021.
---------------------------------------------------------------------------
These models are distinguished by the sizes of their datasets
(extremely large datasets, ranging from hundreds of thousands of audio
samples \9\, to hundreds of millions of images \10\, to billions of
text documents \11\), the amount of computation required to train them
(hundreds to thousands of specialized AI-training computer chips,
running for months), to the complexity of the neural networks (which
now number in the tens to hundreds of billions of parameters).
Foundation Models have already proven to be useful: they can write and
compose code, edit text, produce images, edit images, summarize
documents, form the basis of question-and-answer systems, serve as
potentially useful educational tools, and more.
---------------------------------------------------------------------------
\9\ Introducing Whisper, https://openai.com/blog/whisper/, 2022.
\10\ Revisiting Unreasonable Effectiveness of Data in Deep Learning
Era, https://arxiv.org/abs/1707.02968, 2017.
\11\ An empirical analysis of compute-optimal large language model
training, https://www
.deepmind.com/publications/an-empirical-analysis-of-compute-optimal-
large-language-model-training, 2022.
---------------------------------------------------------------------------
However, the frontier is expensive: it costs millions to tens of
millions of dollars to train these models and therefore they are being
developed by only a small set of predominantly private sector actors. A
challenge we must overcome is how to broaden the experimental
infrastructure necessary to investigate these models, so that more
Americans can participate in the development and analysis of them, and
also learn the engineering and research skills they require.
These examples illustrate how broad the effects of the
industrialization of AI are. But if we want to capture all the upside
of this technology and mitigate its downsides, we also need to think
about policies and investments that can support the burgeoning
ecosystem, and assure the safety and reliability of the systems being
developed within it.
WHY TESTBEDS, DATASETS, AND EVALUATION UNLOCK AI
INNOVATION
While there are many reasons to be optimistic about the potential
opportunities afforded by AI, there are also well-documented risks \12\
and biases \13\ inherent in many of today's applications. These risks
and biases make it harder to deploy safe AI systems, and because these
risks and biases are hard to identify, they can also lead to AI systems
being deployed which have inequitable or harmful behaviors. However, we
can mitigate these issues through pre-deployment and post-deployment
testing, both of which the government can support--specifically, via
the National Institute of Standards and Technology (NIST).
---------------------------------------------------------------------------
\12\ The Malicious Use of Artificial Intelligence: Forecasting,
Prevention, and Mitigation, https://arxiv.org/abs/1802.07228, 2018.
\13\ What Do We Do About the Biases in AI?, https://hbr.org/2019/
10/what-do-we-do-about-the-biases-in-ai, 2019.
---------------------------------------------------------------------------
To maximize the potential of AI technologies, one important role
the government can play is in developing a robust ecosystem for AI
assurance. An assurance ecosystem allows multiple stakeholders to
assess AI systems for performance and safety through a combination of
shared testbeds, datasets, and evaluations. System assurance provides
model developers with certainty in the reliability of their models, end
users with trust that models will act as intended, and government
stakeholders with confidence that systems are safe for the general
public. We can imagine this assurance ecosystem as being analogous to
how product safety standards give consumers confidence in things
ranging from cars, to food, to drugs. It's definitely time to build out
this ecosystem for AI.
Beyond improving the safety and reliability of AI systems, shared
testbeds and evaluations enable a stronger R&D environment. Generally
speaking, whenever a set of researchers create an artificial
intelligence model, they then run that model through a large-scale
battery of tests to assess model performance against previous
iterations, as well as other results in the public domain. External
benchmarks provide an objective, baseline measure from which developers
can compare and improve their systems.
At the same time, sometimes models are found to have capabilities
that their developers did not anticipate, typically through end-users
running novel or unexpected tests.\14\ These tests can reveal both new
capabilities as well as safety issues--for example, when GPT-3 was
released, external users discovered that the system was able to perform
some basic computer programming tasks as well as text-based tasks.
Similarly, a few years ago, external researchers discovered that
commercially deployed facial recognition systems displayed harmful
biases, via a study named ``Gender Shades''.\15\
---------------------------------------------------------------------------
\14\ Predictability and Surprise in Large Generative Models,
https://arxiv.org/abs/2202.07785, pg 4, 2022
\15\ Gender Shades, https://www.media.mit.edu/projects/gender-
shades/overview/, 2022
---------------------------------------------------------------------------
We already have examples of how these kinds of testing
methodologies can be operationalized; following the publication of
Gender Shades, NIST significantly updated its Facial Recognition Vendor
Test (FRVT)\16\ to include fine-grained, granular evaluations which
were also sensitive to the demographic makeup of potential end-users of
the system. This highlights how you can operationalize testing in a way
that improves both the safety of the system (by reducing the likelihood
of deploying unfair systems), and also giving confidence to end-users
of the system that it is going to perform well.
---------------------------------------------------------------------------
\16\ Ongoing Face Recognition Vendor Test (FRVT), 36th edition of
the report, https://pages.nist.gov/frvt/reports/11/frvt_11_report.pdf,
2022.
---------------------------------------------------------------------------
Given the passage of the CHIPs Act and the funding it seeks to
allocate to NIST, we should consider all the ways NIST can play an
expanded role here. What might it look like to identify areas where
industry and academia would benefit from more robust tests and to seek
to create them? How might NIST construct fact-finding teams to identify
some of the areas of greatest ``evaluation need'' and create tests in
response? And can we take the in-development NIST AI Risk Management
Framework (RMF) \17\ and identify specific evaluation methodologies or
tests that we might invest further in, so as to unlock even more
economic innovation and increase the safety of such systems? (In my day
job at the AI research company I am a co-founder of, Anthropic, I spend
a lot of time trying to better evaluate our systems, and I can tell you
that we generally try to incorporate all the tests that exist outside
of Anthropic into our testing framework. We really can't get enough of
them.).
---------------------------------------------------------------------------
\17\ NIST, AI RISK MANAGEMENT FRAMEWORK, https://www.nist.gov/itl/
ai-risk-management-framework, 2022.
---------------------------------------------------------------------------
These examples highlight the value of testing for both economic
expansion, as well as improving the safety and reliability of AI
systems.
WHERE OTHER NATIONS ARE
AI research and development is global. Most data sources tell us
that, outside America, other key countries for AI R&D include the
United Kingdom and, most pertinently for the field of international
competition, China.
China and the U.S. can, in many ways, now be considered at roughly
the same point in AI development. Both countries approach the
development of the technology with different strengths and weaknesses
that stem from their differing political structures, but both host
burgeoning ecosystems of commercial AI companies, and both are
supported by strong academic research infrastructure. The data bears
this out:
For example, while the U.S. leads in the number of global AI
research paper conference citations each year (30 percent, over
15 percent from China in 2021), China continues to lead the
world in the total number of AI publications (journal,
conference, repository combined), with over 60 percent more
than the United States in 2021 (2022 AI Index)
As it relates to patents on AI technologies, China now files
over half of the world's patents (51 percent in 2021). The U.S.
still leads the percentage of granted AI patents globally, but
that percentage has, on average, decreased over the past 7
years while the percentage of granted patents from China has
steadily increased (2022 AI Index).
Beyond the basic metrics of academic publication, there are some
qualitative examples I can share that illustrate how China has begun to
advance its research and development of artificial intelligence.
ImageNet: Chinese teams became increasingly competitive at the
aforementioned ``ImageNet competition'' in the 2010s. One Chinese team
even won an image recognition challenge within that competition several
years ago \18\. This is an extraordinarily competitive competition and
being in the top-3 placed teams was typically considered impressive,
and winning it is a proxy signal for competence. Put plainly: you win
ImageNet by being extraordinarily good at training image recognition
systems.
---------------------------------------------------------------------------
\18\ Large Scale Visual Recognition Challenge 2016 (ILSVRC2016),
https://www.image-net.org/challenges/LSVRC/2016/results.php, 2016.
---------------------------------------------------------------------------
GPT-3: In 2020, an American AI research company called OpenAI (I
worked there at the time) published a paper on a system called GPT-3.
This system was a so-called large language model (LLM). LLMs are
interesting to AI researchers because they are generic AI systems,
capable of classifying and generating arbitrary text, and performing a
broad range of tasks. LLMs are also distinguished by their cost: GPT-3
cost a lot of money to train, and involved using a large number of
training accelerators (in this case, graphical processing units) to
train a single neural network model; it could be considered a frontier
capability due to this expense and complexity.
After publishing the paper about GPT-3 in May 2020 \19\, the first
public replication of the system arrived in a paper in April 2021. The
replication was a system called Pan-Gu and was developed by the Chinese
telecommunications company Huawei \20\. Other replications followed
(HyperCLOVA from Naver in Korea, and after that Jurassic-1-Jumbo from
AI21 Labs in Israel). Notably, two more replications from Chinese labs
followed in Huawei's footsteps, and this year a research group linked
to Tsinghua University released GLM-130B, a GPT-3-style language model
which is currently the best-performing language model \21\ available as
open source--and it's made in China.
---------------------------------------------------------------------------
\19\ Language Models are Few-Shot Learners, https://arxiv.org/abs/
2005.14165, May 2020.
\20\ PanGu-α: Large-scale Autoregressive Pretrained Chinese
Language Models with Auto-parallel Computation, https://arxiv.org/abs/
2104.12369, April 2021.
\21\ For a detailed breakdown of performance, please refer to this
GitHub page for the model: https://github.com/THUDM/GLM-130B, 2022.
---------------------------------------------------------------------------
Re-identification surveillance: China is far ahead of the United
States in the development of surveillance technology. Specifically, we
can look at re-identification; the task of identifying a person in
security camera footage, then being able to see that person via a
different security camera posed at a different angle and use a machine
learning system to figure out it is the same person. This is a powerful
and chilling capability which violates the norms and privacy
protections we have in the United States. However, just because we
wouldn't necessarily adopt a technology ourselves, it's worth noting
when someone else is ahead on a given capability, no matter how
distasteful. In re-identification, recent research shows that Chinese
teams have continually pushed forward the state of the art, and are
responsible for 58 percent of the top papers in the field \22\. Re-
identification requires a combination of large datasets, the
development of large-scale neural networks, and creative algorithm
design. In other words, being good at re-identification means that
you've built a decent AI competency, and it's worth noting that China
is ahead here.
---------------------------------------------------------------------------
\22\ Measuring AI Development A Prototype Methodology to Inform
Policy, https://cset.georgetown.edu/wp-content/uploads/Measuring-AI-
Development.pdf, 2021.
---------------------------------------------------------------------------
HOW THE UNITED STATES CAN SOLIDIFY AI LEADERSHIP
Besides continuing to invest aggressively in fundamental research,
the United States has a couple of strategic policy investments it can
make to bolster its leadership in artificial intelligence. I've already
discussed the importance of making investments in testbeds, datasets,
and evaluation to further unleash U.S. innovation here.
An additional lever we can use is the provisioning of experimental
infrastructure for our academic community. Specifically, we should make
it easier for America's best academic researchers to access
computational power close to that found in industry, so that our
universities can carry out ambitious experiments near the frontier of
AI research; while it's likely industry will continue to define the
frontier due to the increasingly large-scale resources being invested
in model training, we should ensure academia is able to conduct
experiments sufficiently close to it that they can help with the
important work of safety validation and analysis of cutting-edge
systems. This provides both accountability for the private sector, as
well as letting our academic researchers design tools and techniques to
improve systems deployed in the economy. We also need the necessary
infrastructure to build testbeds for these systems so more people can
spend time working out how to unlock their economic opportunities, and
we can use these testbeds to include a much broader group in the
testing and development of AI, which should help us grow the future AI
workforce.
For this reason, Anthropic firmly supports the goals of the
National AI Research Resource (``NAIRR'')\23\--a shared, public
research infrastructure for academic researchers. We view its
establishment as a necessary and excellent long-term investment in
American AI research, as well as a critical resource for supporting the
training and testing of AI systems. Increasing academic access to the
infrastructure necessary to train increasingly resource-intensive
models will build on the long and successful collaboration between
academia and the U.S. government in creating transformative
technologies and advancements across the U.S. economy. (We should also
note that other parts of science already build large-scale experimental
infrastructure, such as the particle physics community.)
---------------------------------------------------------------------------
\23\ THE NATIONAL ARTIFICIAL INTELLIGENCE RESEARCH RESOURCE TASK
FORCE (NAIRRTF), https://www.ai.gov/nairrtf/.
---------------------------------------------------------------------------
From the 1960s until 2010, research shows that academia represented
the majority of large-scale AI experiments. Between the early 2010s and
today, the level of compute required for the largest scale experiments
has increased by more than 300,000x--and the industry-academic balance
has altered, with the vast majority of large-scale results now being
carried out by industry rather than academia \24\. Few academics can
afford the computing and engineering costs required to build and study
large-scale AI models, such as Foundation Models which can cost
millions of dollars to develop, and this is preventing some of our best
researchers from working on problems found at the frontier. Even in
Canada, where the country's advanced research computing (ARC) platform
has allocated increasing quantities of compute to academics since the
mid-2010s, the number of new applications for compute by Canadian
researchers has grown at >10 percent a year--and demand still outstrips
available supply.\25\ Given academia's role in evaluating the safety
and societal impacts of new technologies, we expect resources provided
by the NAIRR will help restore a healthy balance between American
universities and companies in cutting-edge AI research. Testbeds and
other programs funded by the NAIRR will enable AI safety research and
other research agendas in the public interest.
---------------------------------------------------------------------------
\24\ Addendum: Compute used in older headline results, https://
openai.com/blog/ai-and-compute/, 2019.
\25\ 2022 Resource Allocations Competition Results, Digital
Resource Alliance of Canada, https://alliancecan.ca/en/services/
advanced-research-computing/accessing-resources/resource-allocation-
competitions/2022-resource-allocations-competition-results
---------------------------------------------------------------------------
It's reasonable to ask why something like a NAIRR is necessary,
given that AI is being developed and deployed by a large range of
industry actors. After all, we might ask, isn't this a sign that the
government should concentrate its efforts elsewhere? The answer is that
by developing and funding a NAIRR, we're able to build infrastructure
that will naturally serve as a proving ground for some of the ideas
coming out of academia (and perhaps not yet mature enough to be adopted
by industry), as well as creating infrastructure which is highly
complementary with the testbeds NIST is tasked with building as part of
the CHIPS and Science Act. Concretely, we might imagine the NAIRR
serving as a resource for universities to develop large-scale AI
systems, then we can also use the computational power of the NAIRR to
facilitate a broad spread of universities to run testbeds to see how
well we can turn these systems to often neglected problems; improving
the way we manage our farms, building sensing systems to help us
respond to natural disasters, figuring out ways to make our transport
and logistical systems more efficient, and so on. This would also be
directly enabling for another key aspect of CHIPS and Science--the
NSF's new Technology, Innovation, and Partnerships directorate.
We suspect many of the best ideas for how to harness AI are going
to come from universities across America working to solve problems
relevant to their own communities, giving more institutions a role in
assuring that AI systems are safe, reliable, and well-calibrated to the
problems they are designed to solve. Local context is vital to ensure
tools are well-designed. AI logistics systems will best serve the Port
of Gulfport or the Port of Seattle if local port employees and nearby
universities help inform those systems. These tools will underperform
if they are only pressure-tested in distant labs, rather than by
researchers close to local contexts.
Another example is in healthcare: AI shows significant promise to
enhance healthcare, such as applications that aid in diagnostics or
recommending treatments. However, due to differences between hospitals,
such as imaging equipment, procedures, and local population
demographics, AI algorithms that may work well at one hospital can
perform very poorly at another. One solution to this problem involves
more collaborations between universities and their local hospitals to
develop, test, and fine tune models to serve regional needs, which can
be facilitated by testbeds in combination with the NAIRR. The NAIRR can
also help us work on some of the challenges of the governance of
increasingly capable AI systems. By making available experimental
infrastructure for large-scale experimentation, the NAIRR creates an
opportunity to think about how we govern that experimental
infrastructure. Which experiments should get authorized for using a
large amount of NAIRR resources? \26\ How do we test and evaluate the
systems that result from these experiments? \27\ Which people should
participate in the analysis and curation of the datasets which are used
to develop models on the NAIRR? Once a system is developed on the
NAIRR, how might organizations such as NIST help assure the resulting
system for safety? These are all extraordinarily valuable things to
work on in public, rather than in private as is done today by most
industry actors. Beyond enhancing economic competitiveness and safety,
the NAIRR may also help us identify smart, lightweight regulations that
will be fit for the increasingly powerful AI models that will
distinguish this new period of industrialization.
---------------------------------------------------------------------------
\26\ Centre for the Governance of AI Submission to the Request for
Information (RFI) on Implementing Initial Findings and Recommendations
of the NAIRR Task Force, https://www.governance.ai/research-paper/
submission-nairr-task-force, 2022.
\27\ Anthropic response to Request for Information (RFI) on
Implementing the Initial Findings and Recommendations of the National
Artificial Intelligence Research Resource Task Force https://
www.ai.gov/rfi/2022/87-FR-31914/Anthropic-NAIRR-RFI-Response-2022.pdf
---------------------------------------------------------------------------
CONCLUSION
In conclusion, I'd like to thank this committee for its important
role in the CHIPS and Science Act, enthusiastically support full
appropriation for its programs, and recommend that the additional
investments that the bill proposes be made in testbeds, datasets, and
evaluation as a means of unlocking economic innovation and unlocking
revolutionary scientific research.
Additionally, I want to offer strong support for the establishment
of the National AI Research Resource, a national infrastructure built
to facilitate academic AI research that can also help make sure the
United States stays ahead in one of the most important technology
developments we are seeing today. The NAIRR will ensure that American
scientists keep our Nation at the forefront of understanding frontier
compute technology, creating important research and workforce
opportunities and new avenues for economic growth.
Thank you for the chance to speak. I'll be happy to answer any
questions you have about my testimony.\28\
---------------------------------------------------------------------------
\28\ I used an Anthropic 'Foundation Model' to write the
'CONCLUSION' section of this testimony. I added some specific language
and tweaked a couple of words, but other than that, the text is what
the AI generated. I hope this illustrates how these technologies are
already changing how we work today.
The Chair. Do you want to explain that?
Mr. Clark. Yes. I fed the first two pages of a testimony to
an AI language model that Anthropic developed and then it wrote
the last paragraph. And then I added for, and Senators, part
myself to indicate to you it was written by an AI rather than
me.
The Chair. And it compiled a summation from the data points
that had previously been mentioned? What do you think the----
Mr. Clark. It read the text and tried to predict what a
good finishing conclusion would be. I also said, please write
the conclusion. I tend to be polite to it as well.
The Chair. Thank you. Mr. Breckenridge.
Mr. Clark. Top that.
[Laughter.]
Mr. Breckenridge. Right.
[Laughter.]
STATEMENT OF WILLIAM B. (TREY) BRECKENRIDGE III,
DIRECTOR, HIGH PERFORMANCE COMPUTING
COLLABORATORY, MISSISSIPPI STATE UNIVERSITY
Mr. Breckenridge. Chairman Cantwell, Ranking Member Wicker,
and members of the Committee, thank you for the opportunity to
speak today about a topic that is both timely and critical to
our Nation's global competitiveness and security. As many
experts have highlighted, U.S. research investment has fallen
woefully behind our adversaries in areas that are critical to
our national security.
The topic of this hearing are certainly at the top of the
list if we are to sustain and protect our Nation as a world
leader. My perspective comes as Director of the High
Performance Computing Laboratory at Mississippi State
University, where I have served for the past 30 years.
Our high performance computing capacity provides support
for research in artificial intelligence, autonomous vehicles,
cybersecurity, data science, weather modeling, and other areas
of applied research vital to the prosperity of the United
States and the world.
I am extremely proud of the national presence MSU holds in
high performance computing and the advancement of science
enabled by our HPC systems. Mississippi State University has
had a presence on the world's top 500 fastest computers list
since 1996. At its debut in 2019, our Orion supercomputer, a
5.5 petaflops system, operated in partnership with NOAA, ranked
60th in the world and 5th in U.S. academia.
Much of the growth and success that Mississippi State
University has enjoyed in high performance computing can be
directly attributed to the partnerships we have built with
agencies such as NOAA, NSF, Department of Agriculture's
Agricultural Research Service, and the DOD.
With respect to international HPC capability, according to
the top 500 supercomputing sites list, in 2012 the U.S. was
home to more than 50 percent of the world's 500 fastest
computers, while China had less than 14 percent.
Today, the U.S. is home to only 23 percent of the world's
500 fastest computers, while China has increased their share to
over 45 percent. Furthermore, China has significantly increased
HPC funding to match or exceed the U.S. HPC capacity at the
very high end.
The fastest system on the latest top 500 list is located in
the U.S., however many experts are confident that China has
secretly built two systems that rival the performance of the
fastest U.S. system. Simply put, our adversaries are
outspending us in high performance computing. As you well know,
artificial intelligence and blockchain rely heavily on HPC and
the next major advance in computing will most likely occur with
quantum technology.
Quantum holds the potential to solve complex problems that
are unable to be solved with classical computing systems, but
it also presents many challenges that must be overcome with
significant investments. As an example, in the field of
cybersecurity, quantum technology will place much of today's
public key encryption at risk.
The development of robust quantum based encryption
techniques and the use of quantum machine learning to detect
and defeat novel cyber-attacks are crucial. These challenges
are not just in the development and maturation of hardware, but
perhaps more importantly, the creation and training of a
knowledgeable workforce that is able to develop new software
tools and enable access to the broader scientific community
without the need for understanding quantum physics.
The recent passage of the CHIPS and Science Act has been
heralded as a necessary action for the United States to match
investment in these critical technology areas that are being
heavily funded by those in the international community looking
to displace the U.S. as a leader in technology development and
deployment.
We have certainly experienced it recently with the global
chip shortage based on China dependence and we have strong
evidence that their investment in quantum and AI is clearly
outpacing our investment. The CHIPS and Science Act is an
excellent first step to combat this issue.
I would also like to applaud the bill sections related to
the EPSCoR funding and impacts it will have to vastly expand
the talent base in critical fields necessary to remain a global
leader. We have proven that EPSCoR institutions like MSU can be
a national leader in technology fields such as high performance
computing, contributing to the advancement of necessary
technology, while playing a significant role in cultivating the
workforce of tomorrow's innovative leaders.
I offer the following recommendation for the Committee to
consider. To remain globally competitive and protect our
future, investments such as the CHIPS and Science sector are
critical to maintain our national security to mitigate being
outpaced by adversary nations whose primary goal is to relegate
the U.S. to a second tier technology nation.
We must utilize the expertise that exists throughout the
U.S. and grow the technology workforce, not in just existing
U.S. technology centers, but throughout the country. Finally,
Federal, State, and university partnerships will be critical to
addressing these issues and being unified in developing
solutions.
Thank you for the opportunity to testify before you today.
Bipartisan support of technology and innovation is critical to
our competitiveness, and Mississippi State University stands
prepared to be full participants in support of your efforts.
[The prepared statement of Mr. Breckenridge follows:]
Prepared Statement of William B. (Trey) Breckenridge III, Director,
High Performance Computing Collaboratory, Mississippi State University
Chairman Cantwell, Ranking Member Wicker and members of the
committee, thank you for the opportunity to speak today about a topic
that is both timely and critical to our Nation's global competitiveness
and security. As many experts have highlighted, U.S. research
investment has fallen woefully behind our adversaries in areas that are
critical to our national security. The topics of this hearing are
certainly at the top of that list if we are to sustain and protect our
Nation as a world leader.
My perspective comes as the Director of the High Performance
Computing Collaboratory at Mississippi State University where I have
served for the past 30 years. Our high performance computing capacity
provides support for research in artificial intelligence, autonomous
vehicles, cybersecurity, data science, weather modeling and other areas
of applied research vital to the prosperity of the United States and
the world. I am extremely proud of the national presence MSU holds in
high performance computing and the advancement of science enabled by
our HPC systems. Mississippi State University has had a presence on the
world's TOP500 fastest computers list since 1996. At its debut in 2019
our Orion supercomputer, a 5.5 petaFLOPS system operated in partnership
with NOAA, ranked 60th in the world and 5th in U.S. academia. Much of
the growth and success Mississippi State University has enjoyed in high
performance computing can be directly attributed to the partnerships we
have built with agencies such as NOAA, NSF, the Department of
Agriculture's Agriculture Research Service, and the DOD.
With respect to international HPC capability, according to the
TOP500 Supercomputing Sites list, in 2012 the U.S. was home to more
than 50 percent of the world's 500 fastest supercomputers while China
had less than 14 percent. Today the U.S. is home to only 23 percent of
the world's 500 fastest computers while China has increased their share
to over 45 percent. Furthermore, China has significantly increased HPC
funding to match or exceed the U.S. HPC capacity at very high end. The
fastest system on the latest TOP500 list is located in the US, however
many experts are confident that China has secretly built two systems
that rival the performance of the fastest U.S. system. Simply put, our
adversaries are outspending us in high performance computing.
As you well know, Artificial Intelligence and Blockchain rely
heavily on HPC, and the next major advance in computing will most
likely occur with Quantum technology. Quantum holds the potential to
solve complex problems that are unable to be solved with classical
computing systems, but it also presents many challenges that must be
overcome with significant investments. As an example, in the field of
cybersecurity, quantum technology will place much of today's public-key
encryption at risk. The development of robust quantum-based encryption
techniques and the use of quantum machine learning to detect and defeat
novel cyber-attacks are crucial. These challenges are not just in the
development and maturation of hardware, but perhaps more importantly,
the creation and training of a knowledgeable workforce that is able to
develop new software tools and enable access to the broader scientific
community without a need for understanding quantum physics.
The recent passage of the CHIPS and Science Act has been heralded
as a necessary action for the United States to match investment in
these critical technology areas that are being heavily funded by those
in the international community looking to displace the U.S. as a leader
in technology development and deployment. We have certainly experienced
that recently with the global chip shortage based on China dependence,
and we have strong evidence that their investment in quantum and AI is
clearly outpacing our investment. The CHIPS and Science Act is an
excellent first step to combat this issue. I would also like to applaud
the bill sections related to EPSCoR funding and the impacts it will
have to vastly expand the talent base in the critical fields necessary
to remain a global leader. We have proven that EPSCoR institutions like
MSU can be a national leader in technology fields such as high
performance computing, contributing to the advancement of necessary
technology while playing a significant role in cultivating the
workforce for tomorrow's innovative leaders.
I offer the following recommendations for the committee to
consider. To remain globally competitive and protect our future,
investments such as the CHIPS and Science Act are critical to maintain
our national security to mitigate being outpaced by adversary nations
whose primary goal is to relegate the U.S. to a second-tier technology
nation. We must utilize all the expertise that exists throughout the
U.S. and grow the technology workforce, not just in existing U.S.
technology centers but throughout the country. Finally, federal, state
and university partnerships will be critical to addressing these issues
and be unified in developing solutions.
Chairman Cantwell, Ranking Member Wicker, members of the committee,
I thank you for the opportunity to testify before you today. Bipartisan
support of technology and innovation is critical to our competitiveness
and Mississippi State University stands prepared to be full
participants in support of your efforts.
The Chair. Thank you for that testimony. We will now go
remotely to Mr. Steve Lupien, am I saying that correctly? Hold
on a second, Mr. Lupien. Hold on 1 second. We don't have your
audio.
Mr. Lupien. Can you hear me now?
The Chair. Yes. Thank you so much. Thank you.
STATEMENT OF STEVEN C. LUPIEN, DIRECTOR,
UNIVERSITY OF WYOMING CENTER FOR BLOCKCHAIN
AND DIGITAL INNOVATION
Mr. Lupien. Chairman Cantwell, Ranking Member Wicker, and
members of the Committee, thank you for the opportunity to
testify today on the topic of distributed ledger systems and
how the U.S. can maintain our leadership in this nascent
technology. And also thank you for passing the CHIPS and
Science Act.
My name is Steve Lupien. I am a Lecturer in Digital Assets
at the University of Wyoming's College of Business and Director
of the University's Interdisciplinary Center for Blockchain and
Digital Innovation. I specialize in the many ways that digital
assets are disrupting business and financial systems.
In my educational capacity, I am often asked to explain
what blockchains are. And simply put, they are just a new type
of database that allows multiple people to see the same
information at the same time and provide trust that that
information is valid. Blockchains make data unique, and that is
their true power.
Data on a blockchain is immutable, encrypted, yet private,
and allows trust to be written into code and not left to third
party intermediaries. I have been fortunate to work directly in
many of the forward looking regulation that Wyoming has enacted
over the last several years. Currently, Wyoming has passed 30
bills.
The first is a token taxonomy bill defining digital assets
under the law. I believe both the House and Senate are looking
at that presently. We are the first to map digital assets to
existing UCC law, which I think is important to give businesses
certainty in this space. We created a FinTech sandbox to allow
businesses to innovate under regulatory supervision.
And probably the most important thing we did was passed
groundbreaking legislation called the Special Purpose
Depository Institution Bill that creates regulated banks for
both U.S. dollar deposits and digital assets.
Unlike many in this space, I believe that digital assets
should be allowed to flourish under appropriate regulatory
guardrails while allowing latitude for experimentation,
development, and growth. I don't see digital assets as a threat
to our existing financial and business systems, but instead as
an enhancement that will allow the U.S. to maintain our lead in
finance and business long into the future.
However, the industry is receiving mixed signals from
regulators to date, and we need to align on a comprehensive
regulatory framework that removes any uncertainty and allows
businesses the legal clarity they need to experiment and grow
this technology. Digital assets can encompass much more than
just cryptocurrency.
They are being used for highly efficient supply chain
tracking systems, frictionless and speedy payment systems,
smart contracts that allow businesses to automate countless
contractual agreements, and token--and tokenization of physical
assets through nonfungible tokens, NFTs, that not only allow
the tokenization of intellectual property like artwork, but
also innovations such as ESG tokens for energy source
identification, carbon capture, carbon sequestration, and even
environmental habitat protection just to name a few.
The U.S. Government has just begun to fund research
projects in this space, and I encourage you to explore
opportunities to partner with universities more in discovering
how digital assets can bring about financial and business
leadership as well as public good. Please consider the role of
rural universities in this space.
We are often out in front. For example, the University of
Wyoming was the first Division 1 university to offer a degree
program in blockchain, a minor in blockchain, the first to
build and operate an educational Bitcoin mining lab, and the
first to operate a proof of stake staking pool. I am also a
very strong advocate of fostering digital literacy.
My center recently received funding through the Wyoming
Innovation Partnership with seed funding from the American
Rescue Plan to develop educational programs for Wyoming's high
schools and community colleges. The goal of these programs are
to introduce students to the many options available to them in
STEM in general and blockchain in particular.
We have a secondary goal of encouraging more girls and
women and these--to these opportunities as they are presently
underrepresented in the workforce. And I encourage you to look
at how programs such as these can be made nationwide.
Just this month, the U.S. Department of Commerce published
a report titled, Responsible Advancement of U.S. Competitive
and Digital Assets.
In this report, they highlighted several categories that I
support, and I ask you to do so also, including ensuring
effective regulatory approaches for addressing regulatory gaps,
fostering meaningful public-private engagement to ensure that
digital asset stakeholders across multiple business sectors and
Federal departments and agencies can meet regularly to discuss
important issues, and sustain leadership in technological
research and development that will be advanced for activities
such as increased investment in the Government, academic, and
industry led research.
I thank you all for your time and consideration and welcome
any questions you have.
[The prepared statement of Mr. Lupien follows:]
Prepared Statement of Steven C. Lupien, Director, University of Wyoming
Center for Blockchain and Digital Innovation
Applications of Digital Ledger Technology
and How the U.S. Can Better Compete
Chairman Cantwell, Ranking Member Wicker and members of the
committee, thank you for the opportunity to testify today on the topic
of Distributed Ledger Systems and how the U.S. can gain and then
maintain our leadership in this nascent technology. My name is Steven
Lupien, and I am a Lecturer in Digital Assets at the University of
Wyoming's College of Business and Director of the University's
interdisciplinary Center for Blockchain and Digital Innovation (CBDI).
I am an expert in the many ways that digital assets are disrupting
business and financial systems and the co-author of the textbook
Blockchain Fundamentals for Web 3.0 published in August 2022 by the
University of Arkansas Press: Link
In my educational capacity, I am often asked to explain what
blockchain are. Simply put, they are a new type of database that allows
multiple people to see the same information at the same time and
provide trust that the information is valid. Blockchains make data
unique, and that is their true power. Data on a blockchain is
immutable, encrypted yet private, and allows trust to be written in
code, and not left to third-party intermediaries.
I have been directly involved in much of the forward-looking
regulation that Wyoming has enacted, currently 30 Bills since 2017,
that include:
The first Token Taxonomy bill defining digital assets under
the law,
Mapping digital assets to existing UCC Law,
Creating a Fintech Sandbox to allow businesses to innovate
under regulatory supervision, and
The ground-breaking Special Purpose Depository Institutions
(SPDI) bill creating regulated banks for both U.S. dollar
deposits and digital assets.
Unlike many in this space, I believe that digital assets should be
allowed to flourish under appropriate regulatory guardrails while
allowing the latitude for experimentation, development, and growth. I
don't see digital assets as a threat to our existing financial and
business systems, but instead as enhancements that will allow the U.S.
to maintain our lead in finance and business, and as the world's
reserve currency long into the future.
However, the industry is receiving mixed signals from regulators to
date, and we need to align on a comprehensive regulatory framework that
removes any uncertainty and allows businesses the legal clarity that
they need to experiment and grow this technology.
Digital assets encompass much more than just ``cryptocurrency,''
they are being used for highly efficient supply train tracking,
frictionless and speedy payment systems, smart contracts that allow
businesses to automate countless contractual agreements, and the
tokenization of physical assets through non-fungible tokens (NFT's)--
that not only allow the tokenization of intellectual property, like
artwork, but also innovations such as ESG tokens for Energy Source
Identification, Carbon Capture, Carbon Sequestration, and even
environmental habitat protection, to name just a few.
The U.S. government has been slow to fund research projects in this
space, and I encourage you to explore opportunities to partner with
universities in discovering how digital assets can bring about
financial and business leadership as well as public good--please
consider the role of rural universities in this space, we are often out
in front. For example, the University of Wyoming was the first D-1
university to offer a degree program in blockchain (Minor in
Blockchain), the first to build and operate an education Bitcoin mining
lab, and the first to operate a proof-of stake staking pool.
I am also a very strong advocate of fostering digital literacy. My
center recently received funding through the Wyoming Innovation
Partnership (WIP) with seed funding from the American Rescue Plan (ARP)
to develop education programs for Wyoming's high schools and community
colleges. The goal of these program is to introduce students to the
many options available to them in STEM in general and blockchain in
particular. We have a secondary goal of introducing girls and women to
these opportunities as they are presently underrepresented in the
workforce. I encourage you to look at how programs such as these can be
made available nationwide.
The U.S. Department of Commerce published a report this month
titled, RESPONSIBLE ADVANCEMENT OF U.S. COMPETITIVENESS IN DIGITAL
ASSETS. In this report, they highlighted several categories that I
support, and ask that you do also, including:
Ensuring effective regulatory approaches and addressing
regulatory gaps. This will support the development of a healthy
market that nurtures competition and responsible innovation
while safeguarding consumer and investor interests, market
integrity, financial stability, and national security.
Fostering meaningful public-private engagement to ensure
that digital asset stakeholders across multiple business
sectors and Federal departments and agencies can regularly meet
to discuss issues of import to the digital asset sector and
identify areas where coordination may need to occur. Please
include rural universities here, as well.
Sustained leadership in technological research and
development (R&D) that will be advanced through activities such
as increased investment in government, academic, and industry-
led research, workforce development, and digital literacy.
I thank you all for your time and consideration and welcome any
questions you have.
The Chair. Thank you, Mr. Lupien. We will now turn to Dr.
Bob Sutor. Welcome. Thank you for your presence here today.
STATEMENT OF DR. BOB SUTOR, VICE PRESIDENT, CORPORATE
DEVELOPMENT, COLDQUANTA
Dr. Sutor. Thank you. Chairperson Cantwell, Ranking Member
Wicker, and members of the Committee, thank you for the
opportunity to speak to you today on behalf of the ColdQuanta
Corporation.
I will start with the bottom line, quantum technology and
computing will fundamentally and profoundly change how we live
and do business. It can significantly transform our economy,
national security, and the daily lives of Americans. Congress
and the White House have taken bold and important steps to
secure our quantum future through legislation like the CHIPS
and Science Act and the earlier National Quantum Initiative.
We must now accelerate the development of this technology
in parallel with building a robust domestic supply chain and
workforce. Quantum computers are only in the early stages of
development. The number of qubits or quantum bits that known
quantum computers have today range from single digits to
slightly more than 100.
Practical quantum computers will need hundreds of thousands
to millions of qubits for the applications we need. With
quantum computing, we could develop new materials and
substances, including medicines, antibiotics and antiviral
drugs, and design new, more efficient lithium batteries for
transportation.
We could find new alloys for aerospace, automotive, and
military use. I am confident we will get there. Now, if there
is one thing we have learned repeatedly in the nearly 80 years
of the modern computing era, computers get smaller and more
powerful. We put computers and more of them in places we don't
expect. Edge computing works with data close to where it is
created or used.
We could put quantum computers in cell phone towers and
factories. We will want this processing capability on planes,
ships, submarines, and even satellites. To achieve this, the
United States must invest in scaling up the power of quantum
systems while scaling down their size and cost for use at the
edge.
A data center only strategy will leave us vulnerable to not
having the compute resources we need where we need them.
Quantum has many other applications beyond computing,
positioning, navigation, and timing, or PNT, concerns
accurately locating ourselves and moving to where we need to
be.
News Report states that foreign powers spoof and deny GPS
services to confuse and disadvantage their enemies at war.
Quantum inertial sensors, including accelerometers and
gyroscopes and quantum atomic clocks should be able to replace
GPS, prevent spoofing, or act as a backup in case of local and
catastrophic service failure.
For commercial, defense, and intelligence--from commercial,
defense, and intelligence perspectives, quantum sensors could
provide these required, stable, and accurate measurements for
our use on land, at sea, and in space.
Quantum gravity sensors could assist in finding new energy
in mining resources and detect underground facilities not
apparent from visual examination. In what I think is a massive
understatement, quantum is not easy. These systems require
significant investment, scientific progress, engineering
innovation, education, and skills development to bring into
being. We should not wait.
We must secure our domestic quantum supply chain for
necessary enabling components such as lasers and photonic
integrated circuits. Only with a reliable supply chain can the
United States guarantee it can build the computers and sensors
our nations need. All of this will require a significant
quantum workforce, and it will not be limited to those with
doctorates.
We will need trained manufacturing and IT workers, and
software and hardware engineers. There will be many new jobs
and new types of jobs. We must strengthen our education in
computer science, physics, mathematics, and engineering if we
expect to have a national workforce with the necessary skills
to build and use quantum tech.
Quantum companies need support. There is currently a
shortage of Federal assistance to help small quantum companies
transition their promising cutting edge technology now under
development to prototyping, and then to production, scale, and
capability. We must strengthen and accelerate the academia,
Government, commercial collaboration to get practical and
pervasive quantum technologies in the next several years
instead of the next several decades.
We require better procedural and program mechanisms to
navigate the so-called valley of death stage of development,
where we have scattered investment without integration into
deployed systems of record. We need to do more to track,
manage, and coordinate the many individual Federal quantum R&D
projects across the Government.
We must ensure gaps are understood and covered, most
promising technology is fast tracked and well-supported, and
overlaps and duplications are removed. Thank you for the
opportunity to speak with you.
[The prepared statement of Dr. Sutor follows:]
Prepared Statement of Dr. Bob Sutor, Vice President, Corporate
Development, ColdQuanta, Inc.
Chairperson Cantwell, Ranking Member Wicker, and Members of the
Committee, thank you for the opportunity to speak with you today on
behalf of the ColdQuanta corporation regarding our Nation's leadership
position in quantum computing and other emerging quantum technologies.
The bottom line
I will start with the bottom line: quantum technology and computing
will fundamentally and profoundly change how we live and do business.
It has the potential to transform completely our economy, national
security, and the daily lives of all Americans, including discovering
new medicines, engineering better batteries, and creating many new
jobs.
There is a global quantum race happening. We are not the only
country that recognizes the potential of quantum technology. If we do
not make new strategic investments and organize more effectively at the
Federal level to accelerate the domestic development of quantum, we
could lose.
Science, not fiction
Based on my experience in the quantum industry, I want to share
with you how revolutionary quantum information science and technology
are.
The word ``quantum'' may conjure up thoughts of science fiction,
with television shows such as Quantum Leap and movies like Ant-Man and
the Wasp bringing the ``quantum realm'' into popular culture. Luckily
for us, technologies like quantum computing, quantum inertial sensors,
quantum radio-frequency receivers, and atomic clocks are anything but
fiction.
To give you a realistic example of what quantum computing is
capable of, let's consider the field of chemistry. In this area, we
expect to see many impactful quantum computing applications. We base
quantum computing on the principles of quantum mechanics, which
explains the behavior of atoms, electrons, and photons (or particles of
light). Modeling chemical behavior at the atomic level is notoriously
difficult and slow, and it is often necessary to sacrifice the accuracy
of the solution to get an answer in a reasonable amount of time, even
with massive computer resources.
A molecule that most of us are intimately familiar with is
caffeine. Though caffeine has a notable effect on us, it is a small and
relatively simple molecule. Surely we should be able to model on a
computer how caffeine operates on our brain to keep us awake and alert.
Using a classical computer, we represent information with bits, or
zeros and ones. However, to perfectly represent a single caffeine
molecule in a classical computer, we estimate we would need
10\48\ = 1000000000000000000000000000000000000000000000000
bits of information. For comparison, scientists estimate that if
you counted up the entirety of the atoms in the earth, every rock and
human and molecule of air or drop of water, that number is between
10\49\ and 10\50\.
So to model just one caffeine molecule with a classical computer,
we would need an amount of storage comparable to 1 to 10 percent of the
size of the earth. We will never see this with the classical computer
technology we use today.
The promise of quantum computing
Quantum computing can do better. Through the quantum mechanical
properties of ``entanglement'' and ``superposition,'' quantum computing
promises to solve problems that classical computers could never hope to
do in our lifetimes (or our great, great, great grandchildren's
lifetimes). In quantum computing, we use ``qubits'' or ``quantum bits''
instead of bits. We expect to be able to represent caffeine using only
160,000 qubits. With quantum computing, we could develop new medicines,
antibiotics, and antiviral drugs and design new and much more efficient
lithium batteries for transportation. If we could discover new
catalysts for creating fertilizers, we could have far more sustainable
processes supporting agriculture that use much less energy. And we
could find new alloys and materials for aerospace, automotive, and
military use.
However, it is essential to note that quantum computers are in the
early stages of development. How large are the qubit counts of quantum
computers we know of today using public information? These range from
single digits to slightly more than 100. In 2021, a team from the
University of Science and Technology of China announced that they had
built a 62-qubit computer.\1\
---------------------------------------------------------------------------
\1\ https://thequantuminsider.com/2021/05/10/chinese-research-team-
designs-builds-62-qubit-superconducting-quantum-computer/
---------------------------------------------------------------------------
Quantum computing intersects with cybersecurity because it may be
possible someday to break several kinds of encryption methods we use
today. A recent estimate puts the necessary qubit count for a
successful attack at approximately 20 million.\2\ In July, the United
States National Institute of Standards and Technology announced four
new ``quantum-resistant'' cryptographic protocols that should withstand
attack via future quantum computing systems.\3\
---------------------------------------------------------------------------
\2\ https://quantum-journal.org/papers/q-2021-04-15-433/
\3\ https://www.nist.gov/news-events/news/2022/07/nist-announces-
first-four-quantum-resistant-cryptographic-algorithms
---------------------------------------------------------------------------
Taking quantum to the Edge
I believe many people make the mistake of only thinking of
``quantum supercomputers'' living in data centers, taking up a lot of
room, and having significant energy requirements. If there is one thing
we have learned over and over in the nearly 80 years of the modern
computing era, computers get smaller and more powerful. We put more of
them in places we did not expect. Your smartphone might have been a
supercomputer 30 or 40 years ago. We must consider computing and data
at the Edge.
Edge computing works with data close to where it is created or
used. We might put a quantum computer in a cell phone tower or factory.
We will want significant processing capabilities on planes, ships,
submarines, and perhaps even satellites. We cannot always bring
information back into a data center.
In military situations, we may not be connected to centralized
computing resources. We expect to use quantum computers for AI and
optimization problems like logistics. One of the areas of great
interest at the Edge is federated or distributed machine learning for
AI applications.
The United States must invest in scaling up the power of quantum
systems while scaling down their size and cost for use at the Edge. A
``data center-only'' strategy may leave us vulnerable to not having the
compute resources we need where we need them.
The start of the marathon
These systems require significant investment, scientific progress,
engineering innovation, education, and skills development to bring into
being. We should not wait. If we hesitate or under-invest, other
nations could take the lead in creating and using these technologies
for commercial and military applications. Through legislation like the
CHIPS Act and the earlier National Quantum Initiative, Congress and the
White House have already taken important steps to begin to secure our
quantum future. More is needed. We need to accelerate the development
of this technology and do this in parallel with building a robust
domestic supply chain and workforce.
The race is on, but we have a long way to go to perfect usable
quantum computers. We must rapidly scale these systems to make them
usable.
Other types of quantum technology--such as quantum inertial
sensors, quantum radio-frequency receivers, and atomic clocks--are much
closer to becoming fieldable devices. These have the potential to
protect against GPS denial and improve intelligence gathering with high
sensitivity receivers in the next few years, not decades. Investments
in these technologies will feed back into and accelerate our quantum
computing development. My company, ColdQuanta, uses ``cold atom''
technology we have already deployed on the International Space Station
with the Jet Propulsion Laboratory to build our qubits.\4\ We expect
systems based on cold atoms to scale up to the needed range.
---------------------------------------------------------------------------
\4\ https://coldquanta.com/coldquantas-latest-ultracold-technology-
heads-to-the-international-space-station/
---------------------------------------------------------------------------
Quantum sensors for the near term
``Quantum'' has many other applications. Positioning, navigation,
and timing, or ``PNT,'' concerns accurately locating ourselves and
moving to where we need to be. Doesn't GPS, the Global Positioning
System, already give us this? News reports have stated that foreign
powers have spoofed or denied GPS services to confuse and disadvantage
their enemies at war. Quantum inertial sensors, including
accelerometers and gyroscopes, and quantum atomic clocks, should be
able to replace GPS, prevent spoofing, or act as a backup in case of
local and catastrophic service failure.
Commercially, these can benefit transportation and logistics on
land and at sea. From a defense and intelligence perspective, quantum
sensors could provide required, stable, and accurate measurements for
our use on land, at sea, and in space. Quantum gravity sensors could
assist in finding new energy and mining resources and detect
underground facilities not apparent from visual examination.
Work has begun to use quantum sensing as antennae for radio
frequencies such as those used in communications networks and cellular
devices. We believe we can eventually manufacture smaller and more
sensitive receivers that allow us to use more of the radio spectrum.
For example, a ship, plane, or troops on the ground could have more
compact communications systems with broader connectivity. In the U.K.,
British Telecommunications has already started trials of quantum RF
technologies to augment 5G and presumably incorporate into 6G
eventually.\5\
---------------------------------------------------------------------------
\5\ https://newsroom.bt.com/bt-trials-new-quantum-radios-to-boost-
next-generation-5g--iot-networks/
---------------------------------------------------------------------------
While it may seem more straightforward to focus only on computing,
quantum sensors provide the data we will need to incorporate into
processing at the Edge for commercial, intelligence, and military use.
The data has the advantage that it is already encoded in a manner
usable by quantum computers.
Being in the proper format does not mean we can move data where
needed. The United States must invest in quantum interconnect
technology to link quantum computers, sensors, and memory. Without the
interconnects, we will restrict ourselves to building systems that
cannot scale big enough or use data for the practical applications I
have described above.
As I said at the beginning of this statement, quantum refers to
Nature's structure and behavior at the smallest levels. Quantum
technologies operate at the best resolutions possible. There is no
higher resolution to go to than the quantum scale. There is no ``next
time'' to get this right, to invest more aggressively, for the results
and leadership we must have.
The call to action for skills and components
In what I think is a massive understatement, quantum is not easy.
Programming a quantum computer is unlike writing software for a phone,
laptop, cloud server, or supercomputer. Just as we teach classical
programming in high schools today, we must extend curricula to include
quantum. We must strengthen our education in computer science, physics,
mathematics, and engineering if we expect to have a national workforce
with the necessary skills to build and use quantum tech. We have made a
good start at quantum computing education, but this must accelerate. We
must extend physics and engineering education throughout our university
systems to translate the science of quantum sensors into practical and
ubiquitous products.
While we always need some people with the most advanced degrees at
the leading edge, the quantum workforce will not be limited to those
with doctorates. Just as today, we will need trained workers in
manufacturing, I.T., and software and hardware engineering. There will
be many new jobs and types of jobs, and we must have a trained
workforce to fill them.
We must also secure our domestic supply chain. For many quantum
modalities like ColdQuanta's cold atoms, we need increased access to
lasers and rapid evolution of photonic integrated circuits. We must
drive down the cost and size of these components. Only then can the
United States ensure it can build the computers and sensors it needs.
Ensuring successful government, academic, and commercial collaboration
We appreciate your legislative support to reorient Federal spending
and policy priorities to accelerate quantum development. Now is the
time to guarantee we can execute the program. In particular, we require
better procedural and program mechanisms to navigate the so-called
``valley of death'' stage of development, where we have scattered
investment in and development of the pieces without integration into
deployed systems of record.
From where will quantum innovation come? To date, it has grown out
of academia and been prototyped for government agencies but has only
been commercialized by industry to a limited degree. Software is vital
to making the quantum hardware usable, so in May, we acquired
Super.Tech, a startup spun out of the University of Chicago. We must
strengthen and accelerate the academia-government-commercial
collaboration to get practical and pervasive quantum technologies in
the next several years instead of the next several decades.
There is currently a shortage of Federal assistance to help small
quantum companies transition their promising cutting-edge technology
now under development to prototyping and then to production scale and
capability. These are expensive steps that have sidetracked many
promising advances in the past. We can't let that happen to the best-
of-breed quantum projects now, given the stakes of losing to foreign
adversaries. We need to do more to track, manage, and coordinate the
many individual Federal quantum R&D projects across the government. We
must ensure gaps are understood and covered, the most promising
technology is fast-tracked and well supported, and overlaps and
duplications are removed.
Thank you for this opportunity to highlight the quantum
technologies that will be most valuable to the United States and to
offer my recommendations to secure our leadership.
I am happy to answer any questions you may have.
Qualifications--Dr. Bob Sutor
How can I be so sure that quantum will change the world as I
describe it? I am a 40-year computer industry veteran. I have an
undergraduate degree from Harvard and a Ph.D. from Princeton, both in
theoretical mathematics. In 2017, I became part of the leadership team
of the IBM Quantum program after leading the 300-person IBM
Mathematical Sciences Department for six years. I am the author of the
quantum computing book Dancing with Qubits and the classical-quantum
software development textbook Dancing with Python. I am a frequent
keynote speaker at industry conferences. I have been quoted and
interviewed in the Washington Post, the New York Times, the Guardian,
Barron's, USA Today, CNBC, and many other media outlets.
I moved to ColdQuanta in March after 39 years at IBM because I
believe the value of quantum technology extends far beyond the data
center.
At ColdQuanta, Inc., we trace our roots back to a collaboration
between two brilliant minds: Albert Einstein and Satyendra Nath Bose.
In 1924, Bose sent a letter to Einstein and respectfully asked for his
help. This correspondence sparked their uncovering of a new form of
matter, later named the ``Bose-Einstein Condensate (BEC).'' When we
cool atoms to a few millionths of a degree above absolute zero, they
begin to clump together, condense into the lowest accessible quantum
state, and transition from a gas into a BEC. However, this was
speculation, as the technology needed to create BEC wouldn't exist for
another 70 years.
In 1995, Dr. Eric Cornell and Dr. Carl Wieman created the first-
ever BEC in Boulder, Colorado, at JILA--a collaboration between the
University of Colorado at Boulder and the National Institute of
Standards and Technology (NIST). Drs. Cornell and Wieman won the Nobel
Prize in Physics in 2001.
ColdQuanta was co-founded in 2007 in Boulder, CO, by Professor Dana
Anderson, an academic colleague of Drs. Cornell and Wieman. Since then,
the company has built atomic clocks and quantum sensors, often in
response to requests from United States agencies, including NASA,
DARPA, and the Departments of Defense and Energy. Over the last four
years, we have started the development of a quantum computer under the
scientific direction of Professor Mark Saffman at the University of
Wisconsin, Madison. We are a global company focusing on the United
States and the United Kingdom. Our offices are in Colorado, Wisconsin,
Illinois, and Oxford, United Kingdom, with additional employees in New
York, Texas, California, and Florida.
ColdQuanta is committed to delivering broad quantum advantage
through direct research and development of a wide range of quantum
technology in collaboration with government and academia. We have
become, in many ways, the foundation of a quantum supply chain and
ecosystem. We are the only United States manufacturer of many key
components for cold neutral atom and adjacent quantum research. We have
two hundred employees, over 80 of which have PhDs in physics,
engineering, mathematics, and computer science. We have 23 United
States patents and many patent applications we expect to become patents
with high probability in the next 12 months.
The Chair. Thank you. Dr. Sutor. Thank you so much. Dr.
Jones, thank you for being here. I look forward to your
testimony.
STATEMENT OF HENRY L. JONES II, Ph.D., DIRECTOR,
RESEARCH DEVELOPMENT AND SCIENTIFIC
ENTREPRENEURSHIP, UNIVERSITY OF
SOUTHERN MISSISSIPPI
Dr. Jones. Good morning, Chair Cantwell, Ranking Member
Wicker, and members of the Committee. It is an honor to be
invited to testify on this topic today. For the record, I am
Dr. Henry Jones, Director of Research Development and
Scientific Entrepreneurship at the University of Southern
Mississippi.
Our nation has a strong base of institutions like ours that
is a source of enormous competitive advantage, if we make the
most of our diversity. I have experienced the opportunities and
the challenges presented to us by our broad mosaic of a
country. As an entrepreneur and investor, I have been a part of
creating companies in Silicon Valley, in Chicago, and in
Mississippi, and Alabama.
I was educated in the public school system of my small town
of 1,000 in rural South Mississippi and at a public Mississippi
university before moving West to Stanford University to earn a
Ph.D. in aeronautics and astronautics.
I lived in Silicon Valley during the 90s dot com boom,
inspiring me to start my first company. Using science developed
in Mississippi State University lab, our Government's Landsat 7
satellite, and create cutting edge commercial products.
So from the start of my career, I have seen what academia
and Government and industry can do together. What have we
learned along the way? Government has had a tough time keeping
up with the pace of technology. The people and processes of
industry are built for competition with agile, lean, product
market fit, and other concepts promoting quick iterations that
are intensely customer driven.
The National Science Foundation has introduced these
concepts with its I-Corp program that has spread to the
National Institutes of Health and the Department of Energy. And
the Department of Defense created the Hacking for Defense
course at Stanford, which I have been teaching at Southern
Miss.
USM has implemented these innovation approaches with our
partners at NOAA, preparing its 55 petabytes of data for public
customer driven analysis. I commend this committee for
supporting the new NSF Directorate for Technology, Innovation
and Partnerships, or TIPS, which I believe will accelerate the
impact of these and similar programs. Unexpected innovations
come from unconventional connections.
The latest data driven research and innovation is finding
that cross-pollination of ideas from very different fields is
how great leaps forward take place. EPSCoR, the established
program to stimulate competitive research, enables this type of
progress by bringing together universities and their individual
innovators and exposing them to unfamiliar concepts in multiple
ways.
Thank you, Senator Wicker, for your leadership by ensuring
that EPSCoR States will receive an increase in NSF funding,
which was the action needed to support continued geographic and
economic diversity. In our economic system, the most important
resources are customers and capital.
This committee was on target with the CHIPS and Science
Act, in particular the creation of regional tech hubs to
promote the conditions for new concentrations of these
important resources. Thanks to you, Senator Wicker, and your
colleagues here, one-third of the new 8 to 18 new hubs will
include EPSCoR States as a coalition member, and that matters.
I have been a Mississippi tech executive pitching for
capital on Sandhill Road, where those investors expected us to
move to Silicon Valley as we grew. As these tech hubs are
created, special consideration should be given to the alignment
of capital pipelines, from angel to venture, or else the future
tech hub success stories will feel that same gravitational
pull. Diversity is a national resource for resilience.
The tech industry sees the competitive necessity of a
diverse workforce. Conforming cultures lead to groupthink and
being blindsided by unconventional ideas. Companies are seeking
diversity in every dimension it can be achieved, yet are
struggling in the most significant area, enough trained U.S.
citizens.
I am sure we have enough capable U.S. citizens, but we are
missing something in the early stages of the educational
pipeline. There are generational and socioeconomic barriers to
STEM futures that can feel insurmountable. Our university has
recently modified our computer science curriculum to
incorporate industry certifications as milestones within our
degree programs in case a student can't put 4 years of courses
together at once.
Academia has more innovation that we can do. My biggest
fear in regard to securing our leadership in these technologies
is that instead of applying the energy and resources to
accelerate, that we coast instead. Like a race car driver
seeing open road ahead but with no rearview mirror, we won't
know that our competition has passed us until they are out in
front. Even then, it takes time to accelerate and catch up if
we can.
This hearing, the CHIPS Act, EPSCoR, and the NSF TIP
Directorate are the right things to do for us to accelerate
now. Thank you again for this opportunity to speak with you.
[The prepared statement of Dr. Jones follows:]
Prepared Statement of Henry L. Jones II, Ph.D., Director,
Research Development and Scientific Entrepreneurship,
University of Southern Mississippi
Good morning Chair Cantwell, Ranking Member Wicker, and members of
the Committee. It is an honor to be invited to testify on this topic
today. For the record, I am Dr. Henry Jones, Director of Research
Development and Scientific Entrepreneurship at the University of
Southern Mississippi (USM).\1\ Like most of the 3,000 four-year
universities around our country, USM serves local, regional, national
and global constituents through our bright students and dedicated
faculty and staff. Our nation has a strong, diverse base of
institutions like ours that is a source of enormous competitive
advantage--if we make the most of it. If instead our policies promote a
narrow-minded approach that directs attention to a few big brand name
institutions, this national advantage disappears and we lose the
resilience of our country's diversity.
---------------------------------------------------------------------------
\1\ https://www.usm.edu/
---------------------------------------------------------------------------
I have experienced the opportunities and challenges presented to us
by our broad mosaic of a country. As an entrepreneur and investor I
have been a part of creating companies in Silicon Valley and Chicago--
and in Mississippi and Alabama. I was educated in the public-school
system of my small town of 1000 in rural south Mississippi and at a
public Mississippi university--before moving west to Stanford
University to earn a Ph.D. in Aeronautics and Astronautics. I lived in
Silicon Valley during the 90s Dot Com boom, inspiring me to start my
first company using science developed in a Mississippi State University
lab and our government's Landsat 7 satellite to create cutting-edge
commercial products for foresters. From the start of my career, I have
seen what academia and government and industry can do together.
What have we learned along the way?
Government has had a tough time keeping up with the pace of
technology. The people and processes of industry are built for
competition, with Agile, Lean, Product/Market Fit, and other concepts
promoting quick iterations that are intensely customer driven. The
National Science Foundation (NSF) has introduced these concepts with
its I-Corps program \2\ that has spread to the National Institutes of
Health (NIH) and the Department of Energy (DoE), and the National
Security Innovation Network (NSIN) \3\ within the Department of Defense
(DoD) created the Hacking for Defense (H4D) course \4\ at Stanford
which I have been teaching at Southern Miss as Designing Solutions for
Defense (DS4D).\5\ USM has implemented these innovation approaches with
our partners at the Coastal Data Development program within the
National Centers for Environmental Information at NOAA,\6\ preparing
its 55 Petabytes of data for public, customer-driven analysis. I
commend the Committee for supporting the new NSF Directorate for
Technology, Innovation, and Partnerships (TIP),\7\ which I believe will
accelerate the impact of these and similar programs. I was on a call
earlier this week with Jason Calacanis,\8\ one of the most successful
Silicon Valley investors in early stage companies, and he listed areas
like healthcare and hardware where startups struggle greatly for
funding due to bureaucracy and insufficient basic research. I recommend
that TIP's Assistant Director Gianchandani contact Mr. Calacanis and
other startup experts, to ask for suggestions for how the NSF can
change that math.
---------------------------------------------------------------------------
\2\ https://beta.nsf.gov/funding/initiatives/i-corps
\3\ https://www.nsin.mil/
\4\ https://www.nsin.mil/hacking-for-defense/
\5\ http://ds4d.usm.edu
\6\ https://www.ncei.noaa.gov/
\7\ https://beta.nsf.gov/tip/
\8\ https://www.linkedin.com/in/jasoncalacanis/
---------------------------------------------------------------------------
Unexpected innovations come from unconventional connections. The
latest data-driven research in innovation is finding that cross-
pollination of ideas from very different fields is how great leaps
forward take place. EPSCoR,\9\ the Established Program to Stimulate
Competitive Research, enables this type of progress by bringing
together universities and their individual innovators and exposing them
to unfamiliar concepts in multiple ways. Thank you, Senator Wicker, for
your leadership by ensuring that EPSCoR states will receive an increase
in NSF funding, which was the action needed to support continued
geographic and economic diversity. I believe in this program, too--I
serve on a statewide EPSCoR board because it makes sense for our
universities to work together. What I observed is that the real
potential of EPSCoR is its creation of new relationships, better
communications between institutions at the faculty level, and policy
changes that align incentives for working together, all to create a
long-lasting environment of unconventional collaborations.
---------------------------------------------------------------------------
\9\ https://beta.nsf.gov/funding/initiatives/epscor
---------------------------------------------------------------------------
In our economic system, the most important resources are Customers
and Capital. For this reason, a few large urban areas are
understandably attracting more than their share of each. This Committee
was on target with the CHIPS and Science Act, in particular the
creation of Regional Tech Hubs, to promote the conditions for new
concentrations of these resources. Thanks to you, Senator Wicker, and
your colleagues here, one third of the 18 new Hubs will include EPSCoR
states as a coalition member, and that matters. I have been a
Mississippi tech executive pitching for capital on Sand Hill Road,
where those investors expected us to move to Silicon Valley as we grew.
As these Tech Hubs are created, special consideration should be given
to the alignment of capital pipelines, from angel to venture, or else
the future Tech Hub success stories will feel that same gravitational
pull. Similarly, the Hubs should consider the presence of customers as
the driving force for technology adoption, not the location of the
technologists, possibly creating `virtual hubs' for certain types of
problems like cybersecurity where the customer environment is primarily
online.
Diversity is a national resource for resilience. My friends at the
big tech giants like Microsoft and Amazon tell me that those
organizations have seen the competitive necessity of a diverse
workforce. Conforming cultures lead to groupthink and being blindsided
by unconventional ideas. These companies are seeking diversity in every
dimension it can be achieved, yet are struggling in the most
significant area--enough trained U.S. citizens. I am sure we have
enough capable U.S. citizens, but we are missing something in the early
stages of the educational pipeline. One of my childhood friends is a
fourth-generation logger, yet his son is a natural with computer
programming--he's a proud nerd. This family doesn't know what a STEM
career looks like or how to afford a technical education. Luckily we
now have him on a path to maximize his innate skills and interests, but
through this experience I'm discovering that there are generational and
socioeconomic barriers to STEM futures that can feel insurmountable.
Our School of Computing Science and Computer Engineering \10\ at USM
has recently modified our computer science curriculum to incorporate
industry certifications as milestones within our degree programs, in
case a student can't put four years of courses together at once. The
Center for Military Veterans, Service Members, and Families at USM \11\
introduces veterans to resources within higher education, and also
prompts the academic community to make changes to welcome these high
potential students who aren't coming directly from high school.
Academia has more innovation we can do.
---------------------------------------------------------------------------
\10\ https://www.usm.edu/computing-sciences-computer-engineering/
\11\ https://www.usm.edu/military-veterans/
---------------------------------------------------------------------------
My biggest fear in regard to securing our leadership in these
technologies is that instead of applying the energy and resources to
accelerate, that we coast instead. Like a race car driver seeing open
road ahead but with no rear-view mirror, we won't know that our
competition has passed us until they are out in front. Even then, it
takes time to accelerate and catch up--if we can. This hearing, the
CHIPS Act, EPSCoR, and the NSF TIP Directorate are the right thing to
do--we need to accelerate now. Since I started college 30 years ago,
every year I have shared classrooms and workplaces with international
colleagues--hard-working, curious, and intelligent. We are all familiar
with our trade imbalance, but what about our insight imbalance? Do we
have widespread knowledge of what China is doing and can do, in the
same way they know about us? Could one of the programs mentioned above,
or another one, start sending U.S. citizens to China in much larger
numbers to begin to learn from them? That would be the wise action of a
leader who wants to stay ahead. How do we get a very clear image in our
rear-view mirror, and confidently step on the gas?
Thank you again for this opportunity to speak with you today.
The Chair. Thank you, Dr. Jones. We will now go to a round
of questions, and I am going to ask my questions first and then
turn it over to Senator Hickenlooper to chair the rest of the
meeting. And I appreciate his leadership at the Subcommittee
level.
Dr. Albritton, obviously one of the things we need to do to
further our efforts here is the $13 billion that was authorized
for NSF Foundation STEM education efforts. And as we have
passed previous COMPETES Acts, what has happened,
circumstances, you can say the downturn of the economy and we
didn't fully fund the competition bill. So how do we explain to
people the need for these STEM dollars?
And second, you mentioned the areas of NSF expertise you
are already involved in as it relates to these computational
sciences. The bill was all about translational science. It was
about a new directorate to translate science faster. Can you
explain what are the testbed needs of, an example the
University of Washington, to actually help us combine both the
workforce and the infrastructure that is needed to do the
translational science?
Dr. Albritton. Absolutely. I will start with the first
question about, I believe, explaining to the public about our
workforce needs and how we need to fund this, if I have it
right, Senator Cantwell. I think, you know, look at me, I grew
up in Louisiana.
So having the opportunity as a young person to understand
the excitement of science and engineering. If you look at my
career, I have been to ping pong in that area, but make sure we
intentionally tell young people, hey, this is for you, this is
an exciting area, you can have a future. Tell their parents too
this is for their children.
They can grow and become great contributors to the U.S.
economy. So I think starting early but then making the
opportunities available. You know, we hear about college costs,
for example. But as we bring in these more diverse groups of
people, I think there should be funding to make sure that they
succeed and prosper.
So programs that support our students as they come into the
universities, ensure that every one of them graduates, has
interactions with industry so that they can see their future
and prosper. So those are some of the----
The Chair. Well is it safe to tell our appropriator
colleagues who may look at this authorization and then decide
to pass that they are going to fumble the ball?
Dr. Albritton. Yes, they will fumble the ball. And we have
already heard from my colleagues that other countries are out
in front of us. If you look at the numbers of students going
into STEM, it is far lower than we need. In the State of
Washington, we don't have the capacity even for the students
that want to.
We cannot fumble this ball. It is a global competition. We
don't want to lose this competition. And we need everyone
moving forward and playing on this football field, so to speak,
so that we have got all the brainpower in the U.S. moving us
toward the goal post.
So I think yes, we would have fumbled badly if we don't
want to invest, make sure we have good educational support
systems for our students so they can see themselves succeeding
in engineering and science, and then put in the necessary
infrastructure and support systems so that they grow as
students and they actually graduate and prosper.
The Chair. So on the translational science part, what are
the things that we need to get test labs still established?
Dr. Albritton. Yes, so test labs. So our vision at the UW
is to have open, accessible regional hubs. And let's talk about
quantum. Very, very expensive technology. State of the art
systems are cooled to almost absolute zero, very environment
sensitive.
The National Quantum Initiative funded some regional large
areas, but it would be truly great to have qubits for our
students to come and interact with, for example, where they can
learn the quantum foundations, technology. They can just get
really jazzed up because of what they have just done by
flipping a qubit or something.
So we want regional hubs that have state-of-the-art
infrastructure for education, workforce development, even
people that are already out beginning their careers, pulling
them back to reeducate them for this new emerging area. And
then also testbeds that have state-of-the-art characterization
tools, et cetera.
So researchers don't need monumentally huge grants, but
they can go into this regional facility that is easily
accessible and do their wild and crazy ideas, their innovative
thinking. So all of these, I think, on these regional hubs
would really lift the U.S. quantum ability.
The Chair. Well, thank you for that. I think we are going
to have to explain to our colleagues on appropriations exactly
how the other aspects.
There is already some foundation in the bill for us to do
the testbeds, but we have to show them that this is a
combination of both. So, thank you. Again, thank you to Senator
Hickenlooper for chairing the rest of the hearing.
Senator Hickenlooper [presiding]. Great. Now we can really
get started--just kidding. I get the privilege of calling on
the Ranking Member to ask some questions, and just I will
preface that just make sure everyone is aware that he has been
one of the great champions in the U.S. Senate in terms of
maintaining American leadership in innovation and science.
Senator Wicker.
Senator Wicker. Well, thank you very much, Senator
Hickenlooper and Madam Chair. Mr. Breckenridge, tell us--defend
your statement that Mississippi State is a national leader in
high performance computing. And when this EPSCoR expansion
sunsets in 2029, do you have any doubt in your mind that the
leadership of this country, whoever it is at that point, will
conclude that this was an excellent decision and should be made
permanent?
Mr. Breckenridge. Well, as I stated, MSU has been involved
in high performance computing for quite some time. We developed
our first computing cluster system in the late eighties, well
before this technology became the standard that it is today
that most supercomputers are built utilizing.
We have developed the software stacks that enable high
performance computing and message passing interface, software
middleware developed at MSU in partnership with the Department
of Energy and Argonne National Laboratory.
We pioneered many of the technologies that are used for
high performance computing communications, from early in the
90s with Marinette Technologies, these high bandwidth low
latency interconnects, to the standards that are viewed today
in InfiniBand.
MSU has been a leader in that technology. We have also
excelled in the use of high performance computing. It is not
just building these systems and designing them, it is actually
utilizing them to advance science.
So MSU is widely viewed as the leader in grid generation,
and we are doing huge amounts of work today in autonomous
vehicle work and in weather modeling and cybersecurity and so
many other areas that are heavily dependent upon high
performance computing.
Senator Wicker. Thank you. And I just wanted to give you
the opportunity to say to the rest of the country, those people
that are listening and that will be reading the transcript,
that you are a national leader there, in the small town of
Starkville, Mississippi, and not some land grant outpost that
is trying to grab a little Federal money.
Let me turn to Dr. Jones. This EPSCoR, 20 percent goes back
to 13 percent or so in 2029. Are we going to be able to defend
this decision? Any doubt in your mind about that?
Dr. Jones. Yes, sir. I believe in an EPSCoR program like
you do--[technical problems]--I believe, sir, in the EPSCoR
program like you do. So much so that I serve on a statewide
EPSCoR board as an industry member. I think I was chosen
because I have relationships with all of the statewide research
institutions, and I knew that it was imperative that our
universities work together.
And we will see the results of that collaboration, not
necessarily in the technology that is developed because that is
hard to predict, especially on the front end. What we will see
in the long term are new relationships, better ways to
communicate, changes in policy that align incentives across the
universities to work together.
And so consequently, we will have long lasting
collaborations of these new sorts that I mentioned earlier.
Senator Wicker. I just think the results that we produce
out of this increased research across the length and breadth of
our country is going to be recognized. Can you take a few
seconds and tell us about the virtual hubs that you mentioned
in your testimony?
Dr. Jones. Sure. So one of the things that I think that we
can tend to do is we operate like things have always been
instead of thinking about having a vision for how they might
be. And so we could connect people when they are most--they are
closest geographical neighbors, because that is the way that we
would often done it in the past.
But we could connect our EPSCoR neighbors because of the
nature of the expertise that is underlying their ability to put
new technology together, to bring customers to the table, to
pull--to create partnerships, to communicate well.
So a virtual hub might not look like the hubs of the past
in the same way that our response to the pandemic has allowed
us to use virtual connections and to meet online as normal as
it is to meet in person. We can use that same change in mindset
to connect our EPSCoR States around the country.
Senator Wicker. Well, it seems that my time has expired,
but if the Chair would indulge me. Mr. Clark, your virtual
intelligence didn't have to be too smart to conclude that you
were really keen on the National AI Research Resource, but it
did include it in the conclusion. For those of us who didn't
make an A in chemistry and majored in the liberal arts instead,
can you help us understand why that is so important and what it
is?
Mr. Clark. Absolutely, Senator.
Senator Wicker. And minus one minute.
Mr. Clark. Yes. The National AI Research Resource is an
attempt to create a shared infrastructure, a big computer that
academics around America can access to put them on a level
playing field with the private sector.
One of the ways that we hire people at my company,
Anthropic, is we find people from Tier I research universities
and one of the incentives we can offer them to join us is we
have more computers than them.
And I don't think we should necessarily have more computers
than MIT or Stanford or Harvard or really any large scale
research university. I think that that is why you lose
academics, and you lose the teaching base.
So we need to do something to keep these people there and
let them do experiments close to the ones being done in
industry and then we can all learn from their insights.
Senator Wicker. Thank you. And thank you for your
indulgence, Mr. Chairman.
Senator Hickenlooper. Great. Again, I want to thank you
all. Already, I think you all have displayed a lot of the
rationale behind why we so excited to get you all together
here. I want to start with Dr. Sutor, and we are--there are a
number of hubs for quantum computing. Old---CU Boulder and NIST
have had a partnership dating back to the 1960s and put
together the JILA.
Obviously we are going to need--one of the benefits of
those partnerships is helping create the workforce. I think
almost all of you have alluded to the importance of maintaining
a diverse and qualified workforce. How can we expand workforce
and educational opportunities in quantum computing?
We have heard about the expense of creating these hubs. And
what can we look--what can we learn from the success of CU
Boulder's ecosystem in that regard?
Dr. Sutor. Well there are two main points to make there.
And so first starting with Boulder. There was tremendous
insight, as you said, in 1960s between the Government through
NIST and CU Boulder to build JILA, which was actually
originally about astronomy but is now the most precise atomic
clock in the world, is in Boulder, Colorado, developed there.
So a tremendous amount of core science there that spills
over, of course, into the education, many opportunities for
undergraduates and graduates. But then companies come out of
that. There was a celebration yesterday of a startup company
that was founded by people originally at JILA.
So these three axes of academia, Government, eventually
leading to commercial success. Now, the other examples are
Colorado School of Mines, which recently put together a program
on quantum engineering. We hear a lot about quantum information
theory and quantum science.
We should talk much more about quantum engineering because
that is translating it to implementation. It is a
multidisciplinary program. It is mathematics. It is physics. It
is chemistry. It is computer science as well. We have to create
very practical and complete systems that work.
As odd as it may seem, we can distribute our investments,
but ultimately we need working systems, and we need people who
can use them. As I said in my statement, there are going to be
many new types of jobs, jobs that we don't even know the names
of yet. Quantum computing will go beyond classical computing.
So these jobs are not simply replacing what we have
already. We are going to need the tens of thousands of new jobs
on top of everything we have now.
Senator Hickenlooper. All right. Great. I appreciate that.
And I have more questions, but I want to rotate my way--I at
least will be here for a second round of questions. I can
guarantee you all that.
Dr. Clark, experts have said that there are 20--up to 21
different standards of fairness in terms of trying to work
through this in regards to artificial intelligence. Public and
private sector entities often build their custom processes as a
competitive advantage to ensure fairness in a particular
domain.
How can we start beginning to aggregate, to concentrate
around agreed common definitions of what fairness--and fairness
is one of the most closely held moral values among most
Americans? Why is it so important that policymakers start to
align fairness policies?
Mr. Clark. Thank you, Senator. There is a good example
already to emulate the work NIST does today. It has a test
called the facial recognition vendor test, which tests out
facial recognition systems along a whole bunch of axes that
test for fairness and bias.
And what NIST did is took a whole bunch of different
industry standards and approaches in academia and industry and
pushed them together into one test that it worked on in
partnership with a bunch of people.
And we need to do something similar for AI, where we need
to find a way to take all of thinking happening in academia and
industry and find a central place where we can work through
these issues and arrive at a test or set of tests that we feel
is fair and representative.
I would love for that to be a race to the top on fairness
in industry, and I think we can unlock that if we create this
kind of test.
Senator Hickenlooper. You and me both. Let's see, I was
going to have--ask, where was that question? Ah, Dr. Jones,
the--AI powers smartphones, features digital assistants, real
time translation, a lot of those things.
The future of AI is going to need more R&D to incorporate
context relevant to the task at hand. What are some examples--
other examples of functions that AI could better perform if it
could better understand the intentions, the tone of cultural
cues behind the voice command?
Dr. Jones. Well, that is a great question. I think that
maybe, frankly, Mr. Clark might be a better----
[Laughter.]
Senator Hickenlooper. I think that is fair enough. You
know, one thing about a truly expert staff is they will
gravitate toward the right answer for a different question. I
am sure we will come back to Dr. Jones as well.
Mr. Clark. Could you just quickly restate the----
[Laughter.]
Senator Hickenlooper. Asking about the importance of
cultural cues, tone, intentions in terms of what were some of
the examples of functions that AI could do better if, you know,
if it could understand those cues? You know, chime--you know,
voice instructions.
Mr. Clark. Yes. A way to think about this is the reason why
we need more testing by a broader set of Americans is for, as
you build these tests, you learn how to reflect the values of a
greater set of people. AI currently reflects the values of
large amounts of data found on the Internet and some of the
companies that build it, which isn't representative of all
Americans.
And so we can find ways to bring a much larger set of
people into the testing and evaluation of these systems. And
then you can kind of bacon that cultural background knowledge
so these systems are more representative and more cues to the
subtleties of our particular culture.
Senator Hickenlooper. All right. And Dr. Jones, I didn't
know, are you are originally from--you are not from Mississippi
originally?
Dr. Jones. I am actually from, originally from Mississippi.
I had a feeling that Mr. Clark was not.
[Laughter.]
Dr. Jones. Thought he might be best at answering that
question.
Senator Hickenlooper. Well, I thought--again, we will come
back to this. I am out of time, but the--I think the lilt to
your voice which gave away, despite all the Stanford education,
gave away Mississippi origin, that that is one of the big
issues in terms of figuring out how do you get this, some sort
of standardization. Anyway, I will come back on the second
round and now recognize Senator Blackburn who is with us
remotely.
STATEMENT OF HON. MARSHA BLACKBURN,
U.S. SENATOR FROM TENNESSEE
Senator Blackburn. Thank you, Mr. Chairman. And indeed,
another Mississippi voice. And I am glad to see that we have
someone from Mississippi State on the panel today. That is my
alma mater. So good choice there. Mr. Breckenridge, let me come
to you for our first question.
Oak Ridge National Labs in Tennessee, they are home to the
Frontier, which is the fastest computer in the world, and it is
capable of executing 1 quintillion FLOPS per second, and it has
allowed the U.S. to enter the exascale computing era. And this
is unlocking tremendous potential.
We are really so proud of the team that is working on this.
But if you would just touch for a moment what supercomputing
and quantum technologies will unlock, what economic and
benefits that you expect to see coming from this, and then talk
touch on the downsides, whether they are economic or otherwise,
of the U.S. losing the quantum race?
Mr. Breckenridge. Thank you, Senator. Yes, absolutely, the
number one system in the world is at Oak Ridge National
Laboratory, as you indicated. The U.S. must--we must put more
focus into developing and giving access to these systems to a
broader set of our research community.
The systems that are available now to the traditional
university are an order of magnitude smaller or maybe many
orders of magnitude smaller than that of the--at Oak Ridge
National Laboratory.
This is a focus that we have to work on, not only from a
technology standpoint, but from a workforce standpoint as well.
And if you don't mind, would you repeat your second question?
Senator Blackburn. Well, what are the downsides? You know,
I think we know of some of the benefits, and I am pleased that
you mentioned the workforce because since not everybody has
access like the students at University of Tennessee who are
partnered with Oak Ridge and who are working Frontier, I think
that that is one of the downsides. But also, I want you to talk
about what you see is the downsides if we lose this race, this
quantum race.
Mr. Breckenridge. Oh, losing the quantum race has a huge
downsides from a national security standpoint to being able to
do advanced technology. So I think it is something certainly we
are unable to lose. It is just not an option for us.
Senator Blackburn. OK.
Mr. Breckenridge. The--from the benefits of HPC and
quantum, you know, those are many. Whether that is making safer
vehicles or doing a better job in predicting say where
hurricanes may impact us.
This is certainly something that we are facing now. If we
are able to see and develop a better model of where these
hurricanes would make landfall, we could save lives and
increase our responsiveness to that, and to that point,
minimize the economic impacts to us as much as we can.
Senator Blackburn. Well, and I think we all know that
having quantum technologies will open the door for so many
things with logistics, with blockchain, with cryptocurrencies,
with supply chain. Mr. Sutor, I want to come to you on the
supply chain issue. There has been a lot of discussion about
semiconductors, and we have heard about that today.
We have heard about, and we are seeing some of the worst
supply chain issues that we have ever had. So, what--how is
that prohibiting what your company is doing? And how severely
do you see this impacting the growth in this technology sector
for our country as we face these supply chain constraints?
Dr. Sutor. Well, Senator, first of all, so when we speak
about quantum computing, it is an integration of classical
computing as well as this new technology. So any problems we
may have with supply chain, with classical computing, such as
semiconductors, will spill over into our inability to do things
with quantum.
But there are some new technologies such as photonics,
photonic integrated circuits, as well as lasers. With
ColdQuanta, we sometimes say we shine lasers at very tiny
things, and those are being atoms. Well, we need many atoms,
many qubits, ultimately tens of thousands or millions of those.
We need extremely robust laser technology that we can get
when we need it. We have to scale these things down. You can't
have 10,000 lasers just to do a little bit of computing. So we
need advanced technology and development, domestic technology
on these so-called photonic integrated circuits.
Think of semiconductors, but instead of pushing electrons
around, we are pushing around photons of light. Many other
countries have developed photonics centers. I believe the
Netherlands have invested over $1 billion in photonics as well.
Are we happy enough to get these sorts of essential supplies
from outside our borders?
So I believe a systematic examination of the different, not
just quantum computing, but quantum sensing modalities, the
constituent parts are a necessary, identification of where they
come from, and can we provide those ourselves? But for us, it
is photonics, and it is lasers.
Senator Blackburn. Thank you. Thank you, Mr. Chairman.
Senator Hickenlooper. You bet. Thank you. Now let's switch
to Senator Klobuchar, again remotely.
STATEMENT OF HON. AMY KLOBUCHAR,
U.S. SENATOR FROM MINNESOTA
Senator Klobuchar. Thank you very much. Thanks for holding
this hearing. Very important. My state brought to the world
everything from the pacemaker to the Post-it note, and like
every other State, we are facing workforce issues. I know that
this has come up already in the hearing.
Dr. Sutor, in your testimony, you mentioned that the future
quantum workforce will not be limited to those with doctorates,
that we need trained workers in manufacturing, IT, hardware
engineering. What are some of the trained skills we are going
to need in our next generation? And if you could be brief, I
have got a lot of questions.
Dr. Sutor. First, computer science. The best classical
computer scientist or software engineer could be a terrible
quantum computer scientist or engineer. It is really so
radically different that we have to start at the very lowest
levels, and in high school, for example, for this training.
We must make sure that we don't just focus on the sciences,
we focus on the practical. We are going to need to manufacture
these components. These are new techniques, and these are the
types of skills and jobs we are going to have to educate.
Senator Klobuchar. OK. Thank you. Mr. Clark, researchers in
my home state at Mayo in Rochester have partnered with
universities and data science specialist to use artificial
intelligence to process incredibly complex clinical data to
uncover causes of everything from cancer to childbirth options
for expectant moms. In your testimony, you highlight the
significant role artificial intelligence will have in improving
health care. How can we facilitate these partnerships?
Mr. Clark. One of the best ways to facilitate partnerships
here is to take out some of the complications involved in
handling this very sensitive data. And the reason I have
referred to the National AI Research Resource so much is it can
prove to be a place where we can prototype new governance
systems that are both sensitive to HIPPA but make it easier for
a broader set of researchers to access medical data.
You know, bring the data to where a large computational
system is so you can train breakthrough systems. And bring as
many universities and hospitals together in a partnership so
you can make sure that the systems you train are fair and
representative and will work on different populations across
America.
Senator Klobuchar. Very good. I think another area that,
actually Senator Wicker and I have a bill, on this, Precision
Ag Connectivity Act, another area where we are going to see
this, and I know you mentioned it in your testimony, Mr. Clark,
is an Ag with best leveraging technologies to be able to--AI
technologies to be able to do everything from figure out how
much water they need, we could save a whole bunch of water, to
planting and other things.
So I am not going to ask any question on that because I
want to, in my remaining time, move on to a few of the
challenges we have. And you, Mr. Clark, I will ask you this. We
need more transparency when it comes to the recommendation
algorithms that most Americans encounter in their social media
feeds every day.
Senator Coons and Portman and I have teamed up on the
Platform Accountability and Transparency Act to require digital
platforms to give independent, verified researchers access to
data with appropriate privacy protections.
You recommend that the Government invest in the measurement
and monitoring of artificial intelligence development. Can you
tell us what kind of information and resources a researcher
like yourself would need in order to fully evaluate how an AI
system would affect users?
Mr. Clark. Thank you, Senator. The primary thing you need
to do is give them access. You need to give them access to the
system and the ability to run tests on it. And I will give you
just a concrete example. I work with AI systems that are made
of billions of parameters, billions of things that we need to
understand.
Recommender algorithms these days have trillions of
parameters. So the time to invest in this and bring researchers
to this is right now. You need to give access to the system.
You need to let them run experiments, and you need to find a
way that not just companies control access to that, but
researchers are able to, even if they find uncomfortable
things, continue that research so we can have that discussion.
Senator Klobuchar. Very good. And also in your testimony,
you highlight the importance of that shared public research
infrastructure with the National AI Research Resource. How can
supporting and NAIRR is helping create leadership in these
emerging industries?
And again, Senator Lummis and I actually have a bill along
these lines with the social media platforms to get at some of
the work that needs to be done with these companies. Could you
answer that quickly? And then I can--you can move on to my
colleagues.
Mr. Clark. Yes, I will be very brief. In my oral testimony,
I compared some of these modern AI systems to Swiss Army
knives. They are actually more complicated than that. They are
like Swiss Army knives where you don't know all of the things
the knife can do until you have built it and spent a lot of
time testing it.
So the reason why you need a National AI Research Resource
is so researchers can train and develop some of these systems
on par with the opaque ones and industry and work out what all
the capabilities are, and which ones need sort of regulations
and which ones are just very exciting and can be used in the
economy.
Senator Klobuchar. All right. Well, very good. Thank you to
all the witnesses. Appreciate your work.
Senator Hickenlooper. Thank you, Senator Klobuchar. And now
I get to call on our Ranking Member. This actually started
within the Subcommittee, the Space and Science Subcommittee,
which I Chair, but really Co-Chair with Senator Lummis from
Wyoming.
STATEMENT OF HON. CYNTHIA LUMMIS,
U.S. SENATOR FROM WYOMING
Senator Lummis. Well, thank you, Mr. Chairman. This--being
on this committee is like a wicked indulgence, because we get
to ask you questions, some of the most cutting edge, brilliant
people in our country. And as I looked at your bios, I am just
extremely impressed with your capabilities.
Dr. Albritton, I have a question for you that I am going to
start with, and then I want to go to Mr. Lupien, who is at the
University of Wyoming. Dr. Albritton, I am interested to hear
about projects in your specialty that you are working on. You
have such an impressive background.
Dr. Albritton. Are you--me, personally, or the University
of Washington?
[Laughter.]
Senator Lummis. Well, either or both.
Dr. Albritton. So me personally, I am a biomedical
engineer, by the way, but I advocate for the entire college.
And I can give a wonderful example about how AI needs more
cultural context. I work on digestive diseases and the
intestinal track. Google is always offering me digestive aids
and other remedies because it can't distinguish what is
research and what is an actual physical problem.
So that is what I do and a lot of entrepreneurship. But my
major job now is really as chief advocate for the College of
Engineering. And the fabulous part is I get to learn. As you
said, it is wickedly fun. I learn about quantum information. I
learn about great battery storage. We have a new technology.
Senator Cantwell was talking about our test bed facility
called QT3 that we are trying to roll out at the University of
Washington, where anyone can come in and play with a qubit. The
students can.
It is just amazing fun and excitement to learn about other
things like our AI centers, AI for social good, really trying
to make our technology benefit humankind, lead to--allow people
to live better lives. It is just a super fun job, so thank you.
Senator Lummis. Yes. This is an outstanding panel and a
great subject. I want to turn to Mr. Lupien. Thank you for
joining us today and representing some of the cutting edge work
happening at the University of Wyoming, my alma mater.
I want to start by asking you to explain distributed ledger
technology and how it can be used in many different industries
and sectors. And also, could you take us into what you see as
the future of distributed ledger technology?
Mr. Lupien. Well, thank you, Senator. That is a very big
question for a couple of minute answer. Distributed ledger
technology, as I mentioned in my testimony, is really just a
new type of data base. Unlike my colleagues on the panel who
are dealing with high performance computing applications in
quantum and AI, blockchain is actually not a high performance
computing application.
It is an applied application that will touch pretty much
everything that we do in our daily lives, very much like we are
experiencing today with the internet. And where we are going to
see the future, Senator, I believe, is how value is going to be
moved much more easily electronically around the world.
And, you know, we are going to be able to move value almost
instantly and with the settlement finality that doesn't exist
in our existing financial systems today, which are based on,
you know, netting balance sheets to move value. We know we are
going to move actual value. And where I really see the future,
Senator, is how digital assets are going to drive some new
technologies such as the metaverse.
And, you know, within the metaverse, which is the way I
basically describe it is, you know, kind of the Internet on
steroids. It is going to be an immersive use of the internet.
Well, digital assets are going to drive that--drive that
technology. It is how we are going to be paid within the
universe. It is how we are going to buy things within the
metaverse.
And so I think that is the exciting part of where this is
going. But, you know, blockchain technology itself is kind of
in its, you know, entering its bread and butter stage where,
you know, it will be doing a lot of things for consumers and
industry and our financial services industry, where we don't
even realize that that technology is what is driving the back
of the house.
So, I hope that answers your question, Senator Lummis.
Senator Lummis. Thank you. And I have only 5 seconds to
spare, so I will yield back. But if there is another round, I
find this a fascinating panel, and I want to thank you all for
taking the time to be here.
Senator Hickenlooper. I suspect there will be--I know there
will be a second round because I am not leaving. So, Senator
Baldwin.
STATEMENT OF HON. TAMMY BALDWIN,
U.S. SENATOR FROM WISCONSIN
Senator Baldwin. Thank you, Mr. Chairman. It is been
wonderful hearing all my colleagues boast about the incredible
assets in their states. I certainly don't want to pass on my
opportunity to recognize that the University of Wisconsin-
Madison was the first university in the country to establish a
professional master's program in quantum computing.
We also are home in the state to a now Hewlett Packard
Enterprises, Cray, which has been very involved in
supercomputing and exascale supercomputers, and we hope we will
have a role, very prominent role as we imagine and move to
quantum.
So I wanted to start, by the way, recognizing Dr. Sutor
that ColdQuanta has a Madison, Wisconsin office and we are very
pleased to be host of that. What I wanted to start with is,
given the passage of the CHIPS and Science Act, and there are
lots of references to appropriations continuing that drive, I
will say I am one of the appropriator members of the Commerce
committee, so I am listening very carefully to that.
But there is this momentum federally to accelerate quantum
tech transition through the, how can we have it go through the
valley of death, as you were describing in your testimony, and
into commercialization? What sort of--how can we use the
opportunity of this momentum right now?
Dr. Sutor. Let me reiterate, we are very happy to be in
Madison. When ColdQuanta decided to in fact start its quantum
computing effort 4 years ago, we turned to Dr. Mark Saffman in
the physics department.
We have about 25 employees there, many of whom were trained
out at University of Wisconsin as well. So it has been a great
asset for us. When we talk about new technologies, they are
often created by small companies. Yes, some of the big ones
invest sometimes startups.
Unfortunately, 90 percent of all startups fail. And I keep
researching this number year after year, and it keeps coming up
that 90 percent of all startups fail. And so one fundamental
question is, are we willing to let 90 percent of all the great
new technology, particularly in quantum technology, which I
think we can all agree is necessary for our future, are we
willing to have a 10 percent success rate of moving out of the
laboratories, the academic laboratories into commercial
production?
So what we do need from the Government is support for
longer term investment to keep these companies going, to get
them to partner with other smaller companies, to systematically
evolve this technology into something that works. That is a
good thing. But that we can manufacture, which is often a very
different question as well.
So this is that valley of death that I spoke about, to go
from great idea to actually manufacturing and selling hundreds
of thousands, tens of thousands and that is where we need help
from the Government.
Senator Baldwin. Yes. And in a number of these cases, the
Government will be the ultimate consumer, I would imagine.
Think of the various examples that have been given in weather
prediction or defense applications, communications. So you know
that is certainly an opportunity. Let me continue on.
In 2017, China demonstrated an ability to utilize what is
called a ``quantum repeater'' to distribute information across
great distances between Earth and space. This technology will
eventually enable secure transmission of information not
susceptible to eavesdropping. And in reality, China is likely
much more advanced in this space than their published work
might suggest.
Unfortunately, competing with China is not only going to be
just about achieving the next technological breakthrough or
demonstration. We have to ensure that we also have the organic
manufacturing base available to support large scale
commercialization for technologies that have serious security
implications like quantum networking, but also for quantum
sensing that will be key for future GPS and satellite
architectures.
So, Dr. Sutor, what areas of the industrial base do we need
to be focusing on for the raw components and material that we
will need to maintain a future quantum manufacturing edge over
China?
Dr. Sutor. As we have spoken about quantum computing, we
have focused on the processing. So that is like thinking of
your laptop and just thinking of the processor, but not the
data. The data is critical and that is what comes from quantum
sensors. That is what is transmitted in quantum networks.
So for one, I think we need to turn around a lot of our
considerations and look at a data first view of what we are
doing with quantum. In the case of quantum networking, the
repeaters, we need quantum memory, something which we really do
not know how to do very well at all.
So as we go through and look at all the quantum
technologies such as memory, what sorts of quantum technologies
might be best suited for creating memory so we can get the type
of secure communication that you mentioned.
Senator Baldwin. Thank you.
Senator Hickenlooper. Thank you, Senator Baldwin. Now I
have questions--and for those of you who haven't been closely
following the processes of the Senate, Senator Young was, I
think it is fair to say, the principle moving force behind the
CHIPS and Science Act and helped begin it. Would not let it go.
Continued to push it various times. It got tied up. He was
relentless.
STATEMENT OF HON. TODD YOUNG,
U.S. SENATOR FROM INDIANA
Senator Young. Well, thank you, Chairman. That is kind. And
I did want to be here for this panel. I thank all of you for
your work and for your presence here so that we might have an
opportunity to discuss the CHIPS and Science Act, which we
think can significantly further our position as it relates to
American leadership in emerging technologies.
That, of course, will be vital to our national security and
our economic health in the future. One of the key provisions in
the CHIPS and Science Act is the regional tech hub provision.
This will activate underutilized, and in some respects
overlooked, regions around the country and support the
innovation of critical emerging tech areas like quantum and
artificial intelligence and some other key areas.
My state, Indiana, is going to benefit enormously from the
establishment of these tech hubs that will bring together a
wide array of stakeholders. It is really why it is so important
that we not only celebrate the passage of this landmark bill,
and a number are, but continue to focus on its implementation
and fund key provisions such as the regional tech hub
provisions.
Just ask Mr. Clark, sir, why is it important for your mind
that we fully fund the CHIPS and Science Act? And how will it
directly impact the work that you are doing on artificial
intelligence?
Mr. Clark. Thank you, Senator. It will impact me in two
ways. One, it gives the ability to create much more ways to
test and assess AI systems for both economic capabilities and
safety issues. We DIY a lot of our own evals in my company and
we try and make those public, but it is hard to sometimes.
If we can have more evals that are public, it unleashes
more companies to build more stuff using AI and creates a huge
asset. And the second way is that it can allow us to fund more
ambitious infrastructure in the public sector so academics can
do more impactful work.
Senator Young. Thank you for that. Dr. Sutor, I know you
can attest to how the private sector is putting significant
resources and sweat equity into the development of quantum
computing. Many Americans view quantum as it's just another pie
in the sky, Silicon Valley buzz word.
It is never really going to impact their lives in a
positive way. Can you discuss why in fact it is the opposite.
And maybe elaborate on the many ways quantum technology is
going to impact all of our lives moving forward?
Dr. Sutor. Let me give two examples where I think many
Americans, many people around the world are already using
quantum. For people who have ever had a knee or a shoulder
injury and have gotten an MRI, that is quantum technology. It
is higher resolution and usually much safer than X-rays. That
technology was developed originally in the United States going
back to the 1940s.
So people haven't realized they have been doing that. But
even more common is GPS. We have quantum atomic clocks in
satellites right now that not only give us precision location,
and just as we drive as I got here this morning, walked here
this morning, but they give us highly reliable timestamps that
we use in high speed networking situations.
If you go to an ATM and you see the little printout there
of what time it is that comes from GPS, that comes from quantum
as well. So in many ways, quantum is already part of our lives,
and it will become increasingly more so.
Senator Young. Those are powerful examples. And in your
testimony, you note that through the CHIPS and Science Act, in
the National Quantum Initiative, the Federal Government has
already taken some important steps to secure our quantum
future. But as you mention, there is more that needs to be
done. What more can the Federal Government do, what more should
we do to ensure the U.S. wins the global quantum race?
Dr. Sutor. Well, I do believe that it comes down to talent
and it comes down to skills ultimately. There has to be
somebody that develops this technology. There has to be
somebody to use the technology efficiently. So we have to
ensure one way or another, either we can train the people we
have, or we can get them.
Quantum talent is spread around the world, including with
many of our allied nations. For example, the Five Eyes
countries as well. So I do think we have opportunities to look
beyond our borders to get the talent we need to evolve this the
way we must.
Senator Young. That is good. And so my constituents who
might be watching this would be encouraged to know that the
CHIPS and Science Act actually creates some mechanisms and can
accommodate crowding in talent and treasure from our closest
allies and developing things like the field of quantum. So
thank you again all for being here. Chairman.
Senator Hickenlooper. Excellent questions. And I think that
it does bear just a moment of diversion on the issue around the
workforce and immigration. And I think that one of the unique
advantages we have in the rivalry, the competition with China,
with countries all around the globe, is that as the leader of
the free world, we are recognized for the group of freedoms
that we sometimes take for advantage and yet make us a very
attractive destination for highly skilled scientists from all
over the world.
And I think it would be interesting just to take a moment
off and just ask each of you how you look at that, just I want
to make sure that I get you each on the record saying, yes, we
need to look at trying to attract and welcome, really welcome
talent.
And by that, by welcoming, I mean they don't come just for
a year or two. They get to help be part of our ecosystem on a
longer term. Dr. Albritton, why don't you start.
Dr. Albritton. I think we need all the talent that we can
get. I think you can look at any of the universities in the
U.S. and see that the talent comes from across the world, not
just locally or even within the U.S.
Many of the companies we are talking about, their founders
are from out of the U.S. but have found homes in the U.S. So
being an open and a welcoming space, I think gives us a
technological advantage that we can capitalize so that we have
a brain collective in the U.S.
We not only educate our own brainpower, but we also pull it
in. And we will be richer and more diverse, and I think more
productive as a result.
Mr. Clark. As a recently naturalized American citizen, I
really care about this. Forty-three percent of the people at
our company come from outside America. They are the best and
brightest. And the thing that breaks my heart is that we put a
huge amount of resources into giving them a path to stay here
and sometimes it is just difficult and they have to go home. I
wish that we could give a path to everyone.
Senator Hickenlooper. All right. Appreciate that.
Mr. Breckenridge. Yes. And being at a university, we
certainly see a diverse level of students that are here. We
absolutely need the smartest people. And there is quite a
supply of talented individuals in the U.S. even though that we
are not reaching out to. I think there is opportunities for us
to improve the interest and desire to participate in STEM
fields.
Senator Hickenlooper. Yep.
Dr. Sutor. So I would add, in addition too, I would say the
university considerations. ColdQuanta is already an
international company, a 200. We have an office in the UK, in
Oxford. We have terrific skills, so we get to augment what we
do with the University of Colorado, the University of Chicago,
the University of Wisconsin, with Oxford University as well.
So these types of international partnerships so we can
achieve our goals are extremely important. So, yes, by all
means, bring the talent here, but let's use the talent where it
exists to get to where we need to be.
Dr. Jones. Indeed, we have no monopoly on intelligence and
drive, and an inherent desire to be successful and change the
world. And I think, just as you said, Senator, that is an
enormous competitive advantage for us.
And how awesome is it that Mr. Clark is a fellow U.S.
citizen with me to sit here and testify in this location on
this topic, and he is just as wholeheartedly in favor of a
strong America as those of us who were born here.
Senator Hickenlooper. My grandfather used to say oftentimes
the adopted child is more passionate about their parents than
the natural born. Mr. Lupien, do you want to----
Mr. Lupien. Yes. Thank you, Senator. You know, I think I
share my view of my colleagues, technology has made the world a
much smaller place. But I think what we need to do is not only
look to bring individual talent, but by putting forward smart
regulation--companies want to be in the United States. And I
think we want to create a welcoming environment to not only
bring individual talent, but to bring companies here as well.
Senator Hickenlooper. Great. Point well made. Thank you for
that. All right. Now, I am going to shift back to--in case you
didn't notice, we have been rotating Democrat, Republican to
make sure this hearing is bipartisan as much as humanly
possible.
Senator Rosen is one of the few and maybe the only Senator
who has actually created software and probably understands more
about technology than most of us. Senator Rosen.
STATEMENT OF HON. JACKY ROSEN,
U.S. SENATOR FROM NEVADA
Senator Rosen. Oh well, I think there is a lot of smart--
thank you for that. There is a lot of smart people here. I just
wrote computer code for a living, so maybe I do have a certain
knowledge, but you are a scientist as well and there are other
engineers and, but I appreciate that.
And of course, thank you for Chairing, and thank you for
everyone for holding this important--the witnesses for being
here as I am listening to you speak about what we need, the
creative and talented individuals that are going to build our
workforce. People to do the work of the 21st century.
We passed the CHIPS and Science Act. We have the bipartisan
infrastructure law. We are going to need talented, creative
people to do so many things. But today I want to talk about the
talented and creative people that we are going to need to help
work on artificial intelligence applications for cybersecurity,
because according to Acumen Research and Consulting, the global
market for security products with artificial intelligence, well
it is expected to reach over $133 billion by 2030.
I talked about this today. AI technology is rapidly
evolving. It is essential we dedicate research to understanding
the full advantages, what AI can do and of course what it can't
do, and its risk and its applications for cybersecurity. And we
want to be sure all industries have access to this information
and technology so they can responsibly plan--will just speak
here about cyber-attacks, and especially to our AI related
tools.
So, Mr. Clark, how can AI related, or AI, excuse me,
enabled tools, well, they can be used to strengthen our
cybersecurity posture across industries, and then on the same
hand, in what ways does AI fall short as a cybersecurity
solution?
Mr. Clark. Today, AI systems have started to be able to
understand code. They can write and generate code and you could
use that to--in the same way I could spot a sentence that has
poor grammar using a line and a language AI, I could spot code
that has bugs in it using a code AI.
So we are right at the beginning of really, really exciting
and useful capabilities here. However, when I wrote my oral
testimony using a language AI, I checked what it wrote before I
said it to the assembled Senators.
And so these kind of applications for cyber-defense are
going to require people working with systems in partnership and
really, really vigorous and well-funded testing to find out if
these systems can be truly relied upon.
Because in the cyber context, you absolutely don't want
something to, for lack of a term, wing it. Thank you.
Senator Rosen. Right. No, I agree. Computers are very good
at either/or. It is this or not this. It is all the algorithms.
It is greater than, less than--they are binary. And so I think
you are right where we have to have the human component and
that is going to get us to the human component built upon STEM
and cyber workforce shortage.
We know we have to grow our STEM workforce, our cyber
workforce across the board. We have 700,000 positions that are
open across the country. I have bipartisan legislation with
Senator Blackburn to invest in cybersecurity apprenticeships.
Senator Capito and I have launched the Women in STEM Caucus
to bring more women into these wonderfully creative STEM jobs.
Worked on a world bipartisan STEM Education Act.
Of course we know the CHIPS and Science Act. But there is
just so much more to do. So, Dr. Albritton, how can the Federal
Government work better with, I would say, creating maybe
apprenticeship programs perhaps, certificate programs, 2 year,
4 year degrees across the spectrum?
We need all kinds of folks in this realm. How do you work
better to potentiate and create this creative new workforce we
are going to need and bring in girls and underrepresented
communities?
Dr. Albritton. Absolutely. So I think one of the things we
need is long term, stable investment and building the number of
educators we have and the diversity of educators that work with
our students to show them the excitement.
I think investment--we can't always think about the next
six term pay off, but what will pay off in 5 and 10 years. And
I often think when we think about education, we are starting at
an early age, maybe high school, maybe elementary school.
So I think it is having staying power and being willing to
invest over time so that you can reap the rewards at some later
time. And a lot of research is like that. So patience, I guess,
is one of the best ways to do it and invest long term--long
term, stable investment in education.
Senator Rosen. I couldn't agree more. I think that when you
excite kids in elementary school about learning about robotics
or computing or science in general, even growing plants,
whatever it is, they can learn about biology, geology,
computers, and they learn to internalize that at a young age.
Very important. Thank you, Mr. Chairman. I yield back.
Senator Hickenlooper. Great. And I think we have a couple
of Senators coming, but since they are not here, I will just go
ahead and begin with my second round, and I hope I can slip in
a question or two. Mr. Breckenridge, you lead a center with
high performance computing.
Obviously, understands a lot of the--understand a lot of
the limitations to how we make sure that these technologies
have broad and equitable access across the country. You talked
about that a little bit. Several people have talked about that
but you as well in your opening statement.
That--and the needs of the diverse workforce. How is access
to supercomputing, as it is just now reaching more research
institutions, how can we promote that equitable access and make
sure that emerging technologies like quantum computing are
getting the maximum benefit from all over the country in terms
of how we build out the engineering that Dr. Sutor talked
about?
Mr. Breckenridge. Sure. So high performance computing is
becoming ubiquitous. Most research universities, maybe even all
research universities, have high performance computing assets
there. They may be large, they may be small, but it gives them
the ability to utilize those assets.
I think we have got to do the same thing with emerging
technologies, whether that is quantum or other areas that
allows the faculty member, the researcher, the student to think
of new and novel techniques and technologies. And without
having that immediate access, it is really limiting.
Senator Hickenlooper. Right. Yes, I couldn't agree more.
Now I will--I am not done with my second round, I am just
taking a short hiatus. Senator Peters.
STATEMENT OF HON. GARY PETERS,
U.S. SENATOR FROM MICHIGAN
Senator Peters. Thank you, Chairman Hickenlooper. Thank you
for taking a little hiatus for me. Mr. Clark, in your written
testimony, you discuss the importance of prioritizing the
development of resources for testing, evaluation, and
benchmarking of artificial intelligence systems.
In Michigan, we are the home of the auto industry and have
been investing significant resources into AI systems to pilot
autonomous vehicles that will be able to drive through complex
city environments without a human driver. But the secret to all
of that is AI systems that are able to pilot these automobiles
and handle the massive amount of data coming in from all of the
sensors.
Now, it has been described to me that self-driving cars is
basically the moonshot for AI. When AI was able to do that, AI
has really proven itself to be able to do all sorts of things
that will transform industry in meaningful ways. So I would
like you to discuss some of the existing barriers to testing,
evaluating, and benchmarking of AI, and what resources do you
think we will need to overcome these obstacles?
Mr. Clark. Thank you, Senator. This is a matter close to my
personal interest. An area that is crucial to doing self-
driving cars well is very large-scale high fidelity simulators.
You need to simulate not just the cities that you are
trying to drive your cars around, but you also need to simulate
people in the cities, weather conditions, all kinds of
variation and that takes resources, and specifically it takes
computational power.
So for more of what we can give researchers access to large
amounts of computation to run high fidelity simulations, the
better we can train these vehicles. That is a fundamental
block.
Senator Peters. Well, thank you. And another question for
you, Mr. Clark. While we know that AI can deliver some
substantial benefits to society, I am also concerned that AI
will create some significant new dynamics when it comes to
cybersecurity.
Providing certainly increased speed and agility for both
our network defenders, but unfortunately also for our
adversaries when it comes to cybersecurity. So how should we be
thinking ahead about ways to continue to improve cybersecurity
as advances in AI continue to march forward?
Mr. Clark. We can take AI systems and pair them with human
experts who will kind of teach the AI systems about how to
secure cyber infrastructure. And you can think of it as like
creating a digital immune system for our networks.
AI model that constantly watches networks and watches for
suspicious activity and tries to actively heal them and repair
them. Additionally, we do need to invest in analysis of what is
going on abroad. Other groups are harnessing this stuff for not
just defense, but offensive purposes.
And if we don't have a very clear and calibrated sense of
where those capabilities are, we could get surprised, which is
not what any of us want.
Senator Peters. No, absolutely not. We need to lean forward
aggressively in that area. I appreciate that answer.
Dr. Sutor, in May, the Biden Administration issued a
national security memorandum requiring more proactive
engagement with critical infrastructure owners and operators to
ease the migration of our computer systems from today's
standard encryption algorithms to quantum resistance
cryptography.
So my question to you is, what do you see as the key
challenges for the private sector in this migration?
Dr. Sutor. I think even if we don't mention the word
quantum, it can take up to 10 years, for example, for a bank to
move to a new cryptographic protocol. It is a lot of work. It
touches many different parts of their systems. So even if
quantum computers today cannot break encryption, if they can do
it in 10 or 15 or 20 years, it is a significant problem.
And so it is something that people have to look at and do a
real evaluation. What sort of cryptographic profiles,
techniques do they use today? Are they RSA? Are they elliptic
curve, for example? When might those be broken and what should
they put in place to be safe about that?
I might also add there are export control issues concerning
this, which is, where do we think quantum computers might be
able to break crypto? And what should we do about that with
respect to export controls?
There has been some very solid work there to define the
sweet spot to maintain competition developments of the
technologies while still protect national security.
Senator Peters. Right. Thank you. Thank you, Chairman
Hickenlooper.
Senator Hickenlooper. Great. I am going to--I have been
told that they are holding a vote open for me on the Senate
floor. I am the last person to vote. So I am going to ask one
last question of Dr. Albritton, and I am not going to get to
come hang out, which is usually my favorite part.
Once we have gotten all this out on the table, then chew it
over for another 10 or 15 minutes. As you try to leave, I would
normally restrain you. At least do my best. Dr. Albritton, we
talked about fairness a little bit with AI, trustworthiness, I
think, is another issue that we keep hearing about more and
more.
The trustworthy AI systems are more fair, more transparent,
more explainable, more robust, I think in many cases. As
requirements for AI systems become stricter, what advantages
could American companies see if they continue--if we continue
to make sure we lead the world in making AI trustworthy?
Dr. Albritton. I think the advantages are unbounded. I
mean, you could go and be confident that no matter what your
race or sex, you will--the facial recognition programs will
treat you equivalent to everybody else. When you apply for a
job, you will get the best candidate.
You will not have a biased AI program, maybe selecting for
certain attributes. So I think it is a more equitable, a more
diverse, and a more economically productive world, if we can do
that, because we will have everyone playing on a level playing
field and everyone will have a chance.
It is a world now, we have heard about all the great
competition. We have got to have everyone having a chance, no
matter what your background, what your other demographics are,
your economics, et cetera. And I think a fair, more fair AI
will give us all those things we need.
So economically, the U.S. is far, far better off with
fairer systems, in my mind.
Senator Hickenlooper. Well, I couldn't agree more. And I
think you have all in different ways played a role in creating
that, the constantly improving sense of fairness and
trustworthiness.
Thank you each, Dr. Albritton, Mr. Clark, Mr. Breckenridge,
Dr. Sutor, Dr. Jones, Mr. Lupien. I think I have questions for
each of you that will go into the written--put them in writing
to you. I guess this will then end today. You are off the hook
now. Conclude our hearing for today.
I would like to thank all my Senate colleagues and
especially our witnesses for this, I think truly fascinating
discussion. I think the best hearings are the ones where you
come away from it and you have got more thoughts in your head
than you came in with, and I think that is certainly true to,
maximum effect today.
The hearing record will remain open for questions for two
weeks until October 13, 2022. Any Senator who would like to
submit questions for the record may do so. Until then, we ask
that witnesses submit their responses to those questions by
October 27, 2022.
There will be harsh penalties--no, I am just kidding. With
that, this committee is adjourned.
[Whereupon, at 12:05 p.m., the hearing was adjourned.]
A P P E N D I X
Response to Written Questions Submitted by Hon. Maria Cantwell to
Dr. Nancy Albritton
Compute Access: Academic institutions represented the majority of
large-scale AI experiments from the 1960s until 2010. U.S. News and
World Report (2022) ranks the University of Washington (UW) as one of
the Top 5 Artificial Intelligence Programs \1\ in the Nation. In July
2021, UW won a $20 million dollar grant over 5 years from the National
Science Foundation (NSF) to establish a new artificial-intelligence
research institute for dynamic systems. However, over the past decade
the industry-academic balance has altered, with the vast majority of
large-scale AI research carried out by industry. Building large-scale
AI models can cost tens of millions of dollars and require 300,000x
more computing power than experiments conducted in the prior decade.
---------------------------------------------------------------------------
\1\ https://www.usnews.com/best-graduate-schools/top-science-
schools/artificial-intelligence-rankings
Question 1. What are the specific gaps you are seeing in the
availability of AI computing resources in industry, government, and
academia? Besides the National AI Research Resource, what else can the
U.S. government do help academia remain competitive, or better
collaborate with, private industry in AI research development?
Answer. Thank you for your leadership of and investment in science,
technology, engineering and mathematics through the development and
passage of the CHIPS and Science Act. Artificial intelligence and
machine learning have spurred significant scientific breakthroughs in
healthcare, biology, and materials discovery. For example, DeepMind's
solving of decades-long challenge of protein folding; the use of
machine learning for more effective chemical synthesis; and PostEra's
accelerated development of COVID-related drug discovery. More resources
invested in emerging technologies will generate sizable social and
economic benefits.
At the same time, we need more investment in AI policy and national
strategies to address ethical concerns (equity, fairness, etc.)
associated with using AI/ML. There are very limited efforts to address
these concerns in the industry. Emerging connections between AI and
quantum computing are advancing rapidly. For example, quantum computing
provides training sets for AI, which then learns how the quantum world
behaves and can make predictions without solving the exact quantum
problem each time. Increased Federal investment in quantum computing
will be critical to maintain our global competitiveness. To continue to
accelerate discovery sustained and increased funding of the Federal
agencies that enable us to remain a quantum information leader in a
fierce global landscape is required.
AI Skills Gap: In a recent survey in the United States and United
Kingdom, 93 percent of surveyed companies reported AI as a strategic
business priority; however, 51 percent found that they lack the right
mix of skills to fully realize those strategies.\2\ As the Frank &
Julie Jungers Dean of Engineering, you serve as the Chief Academic
Officer of the College, overseeing $159 million in research
expenditures and providing leaderships to over 279 faculty and more
then 8,000 students.
---------------------------------------------------------------------------
\2\ https://www.snaplogic.com/company/media/press-releases/ai-
skills-shortage-research
Question 2. How do you recommend the U.S. narrow this AI-skills gap
in industry and, specifically, what can educational institutions like
UW do build a robust and inclusive pipeline of students to meet the
growing demand?
Answer. Universities can play an important role in narrowing the
AI-skills gap in industry. At the UW we are in the midst of
diversifying our curriculum to ensure that all students have
foundational skills in AI and machine learning. For example, in the EU,
the vast majority of specialized AI academic offerings are taught at
the master's level; robotics and automation are by far the most
frequently taught course in the specialized bachelor's and master's
programs, while machine learning (ML) dominates in the specialized
short courses.
In addition, we must advance programs to promote diversity in AI.
Currently, female graduates of AI PhD programs in North America have
accounted for less than 18 percent of all PhD graduates on average,
according to an annual survey from the Computing Research Association
(CRA). The CRA survey shows that in 2019, among new U.S. resident AI
PhD graduates, 45 percent were white, while 22.4 percent were Asian,
3.2 percent were Hispanic, and 2.4 percent were African American.
Failing to prepare our citizens for the innovation economy
compromises our Nation's long-term competitiveness and economic
stability and disadvantages our citizens and communities. Industry,
government and universities must step up and invest in the STEM
workforce. As we enter an increasingly specialized economy, America's
leading research universities, including the University of Washington,
are uniquely poised to provide leadership, research, and workforce
education to meet this need, but we need Federal investment to support
these endeavors.
STEM Training: You mentioned in your testimony that there is fierce
competition to hire STEM faculty to educate students. You also
mentioned that the University of Washington has turned away excellent
students because student demand exceeds the capacity of your STEM
programs. The CHIPS and Science Act authorized $13 billion for STEM
education to help close the gaps in training the STEM workforce.
Question 3. What types and levels of Federal funding do you think
might be most effective at helping close the STEM training gap? As
possible, please provide information on the roles of different types of
funding (e.g., scholarships, research funding, infrastructure support,
and teacher professional development and hiring support) and on
suggested funding agencies, as well as information on how many
additional students could result from each.
Answer. Investment in the Nation's STEM educational system is
essential to grow our economy while preparing our students for the
fields of tomorrow. STEM graduates drive innovation in health, clean
energy, technology, aerospace and infrastructure, but our state's
industries turn elsewhere for workers because our universities cannot
keep up with demand due to the lack of faculty, space and resources.
Insufficient STEM capacity denies our state's students access to high-
impact career opportunities. This inaccessibility is particularly acute
for students traditionally underrepresented in these fields.
To address the STEM training gap effectively, we need the
government at the Federal and state level, industry and academia all
working together. Continued investment in our Federal agencies such as
NIH, NSF, DOE, is critical to these efforts as well as funding for key
programs like Pell Grants. In the last academic year close to 11,000 UW
undergraduates received Pell Grants, more than all the Ivy Leagues
schools combined. The UW awards $100 million a year in institutional
grants and scholarships to Washington students to help them pursue
their education. A third of our undergraduates are the first in their
families to attend college. We receive more Federal research dollars
than any other public university in the Nation. Research funding pays
for graduate trainees as well as undergraduate research experiences.
More than 7,000 undergraduates conduct research per year at the
University of Washington. Core facilities and instrumentation dedicated
to training and research is often very expensive and shared resources
will enable access for all to state of the art equipment both in
academia and industry. Finally, STEM engagement, education and
experiences at all high and middle school levels is foundational to our
future success.
Thank you for the opportunity to participate in this important
hearing.
______
Response to Written Questions Submitted by Hon. Kyrsten Sinema to
Dr. Nancy Albritton
Drought and Fire Monitoring at NOAA: Fire season began in late
April in Arizona, as the Southwest experiences the most severe drought
in twelve hundred years. The Tunnel Fire north of Flagstaff burned over
twenty thousand acres while the Crooks Fire consumed over six thousand
acres south of Prescott. After the fires were extinguished, burn scars
have led to significant flooding issues in Flagstaff and other areas in
northern Arizona.
Arizonans rely on National Weather Service predictions and National
Oceanic and Atmospheric Administration (NOAA) data to predict fire
trajectory and to determine impacts to landscapes and human health. It
is important that NOAA and the Weather Service have the resources and
expertise needed to predict weather and drought conditions.
Earlier this year, NOAA launched two new supercomputers, including
the ``Cactus'' supercomputer in Phoenix, Arizona, which operates at a
speed of 12.1 petaflops.
Question 1. The bipartisan infrastructure law included $80 million
dollars for NOAA high-performance computing technology and $50 million
for wildfire prediction, detection, and forecasting. How will these
resources allow NOAA to improve drought and wildfire prediction
forecasting?
Answer. A record number of our citizens experienced extreme weather
in this past year. These significant Federal investments will enable
NOAA to procure research-supercomputing equipment used for weather and
climate model development to improve drought, flood, and wildfire
prediction, detection, and forecasting. The high performing
supercomputing infrastructure is necessary for the long-term benefits
of revealing the underlying correlation between climate change,
environmental impact and natural disasters, hence contributing to our
ability to understand the complex and large-scale ecosystem.
NOAA plays a critical role supporting federal, state, local and
tribal partners in preparing for the threat of wildfires and in
battling the blazes that endanger life and property. NOAA's forecast
products range from short-term warnings to long-term seasonal
predictions, and include air quality and smoke forecasts related to
wildfires. NOAA also provides real-time fire and smoke detection tools
using new imaging capabilities from geostationary and polar orbiting
satellites. The recently launched JPS-2 satellite will support
essential forecasts for extreme weather events, feed daily weather
models, and monitor climate change. Weather forecasters, first
responders, pilots, climate scientists and fire crews will use the data
collected by the satellites.
STEM Education in CHIPS and Science Act: The bipartisan CHIPS and
Science Act authorized $13 billion over five years for the National
Science Foundation (NSF) to put towards Science, Technology,
Engineering, and Mathematics (STEM) education. The legislation directs
the NSF to explore opportunities to engage students from groups
historically underrepresented in STEM fields as well as rural areas to
expand the pipeline to help ensure the United States has the STEM
workforce we need for the Twenty-First Century.
Question 2. How will the CHIPS and Science legislation expand STEM
education and, in turn, America's STEM workforce? Why is it important
for Congress to fund the investments in STEM education made in the
legislation?
Answer. Investment in our Nation's STEM educational system is
essential to maintain and grow our economy. STEM graduates drive
innovation in health, clean energy, technology, aerospace and
infrastructure, but our states' industries often turn elsewhere for
workers because our universities cannot keep up with demand due to the
lack of faculty, space and resources. Insufficient STEM capacity denies
our citizens access to high-impact and high-earning careers. This
inaccessibility is particularly acute for individuals traditionally
underrepresented in these fields.
A thriving economy demands a well-rounded workforce for long-term
competitiveness. Failing to prepare our citizens for the work of the
future compromises our country's long-term competitiveness, security
and economic stability. To remain competitive on a global stage we must
increase the STEM pipeline, its diversity and leverage all of our
talent.
Investment in America's leading research universities through the
CHIPS and Science Act will allow talented faculty and students to
further innovative science that will elevate the U.S. as a global
destination for knowledge and discovery in quantum information
sciences. These foundational investments will influence economic and
national security, prepare U.S. students for jobs with quantum
information technology, enhance STEM education at all levels, and
accelerate exploration of quantum information frontiers, all while
expanding and diversifying the talent pool for the industries of the
future across the Nation.
Semiconductor Investments: Once the global leader, America now only
accounts for approximately 12 percent of the world's semiconductor
manufacturing. Recognizing this important national security issue, I
worked with my colleagues to ensure the CHIPS and Science Act also
appropriated $52 billion for the Commerce Department to establish
grants to jump-start investments in America's domestic semiconductor
industry. These investments will support tens of thousands of Arizona
jobs at facilities managed by Intel, TSMC, and other Arizona
semiconductor companies.
Question 3. How does the CHIPS and Science Act and the law's
semiconductor investments illustrate the United States' commitment to
being a global leader in this field and what further actions can be
done to expand U.S. leadership in this field?
Answer. Recognizing that the U.S. semiconductor industry's
dependence on offshore manufacturing poses a threat to the Nation's
long-term economic competitiveness and national security, the passage
of the CHIPS and Science Act was critical to the development of new
programs to promote research, development and fabrication of
semiconductors within the U.S.
In the past decades, Institutes of Higher Education in the U.S.
have lost footing in semiconductor design, engineering and
manufacturing curricula. This is because the majority of semiconductor
manufacturing has been produced outside of the U.S. and because
industry has been leading the field both in personnel as well as in
technology.
The technologies and processes required to design and manufacture
semiconductor chips continue to become more sophisticated, requiring a
workforce with a wider and more flexible range of STEM skills. Today
significant gaps exist between the industry-oriented skills in demand
and the traditional skills being supplied. A 2017 survey conducted by
Deloitte and SEMI [3] found that 82 percent of semiconductor industry
executives reported a shortage of qualified job candidates, and the
challenges of finding qualified workers have increased since then, with
gaps at all skill and education levels--from technicians to doctoral-
level engineers. The workforce needs of the industry are expected to
more than double as the CHIPS Act manufacturing incentives in the U.S.
will create tens of thousands of new jobs within the next few years,
concentrated in technical and engineering roles.
The CHIPS and Science Act is a significant investment in new
programs to promote the research, development, and fabrication of
semiconductors within the U.S. I ask that you continue to accelerate
discovery through sustained and increased funding of the Federal
agencies that enable us to remain a STEM leader on a fierce global
landscape. Specifically, continued and additional investment to secure
U.S. leadership in microelectronics manufacturing is needed. In
addition, the establishment of a national network for microelectronics
education that is open to all U.S. universities and colleges to
participate in and that industry and government stakeholders can
collaborate.
[1] https://www2.deloitte.com/us/en/insights/industry/technology/
growing-semiconductor-market.html
[2] precedenceresearch.com/semiconductor-market
[3] https://www.semi.org/en/workforce-development/diversity-
programs/deloitte-study
______
Response to Written Questions Submitted by Hon. Raphael Warnock to
Dr. Nancy Albritton
Workforce Diversity:
Question 1. What is the importance of fostering greater diversity
within America's science, technology, engineering, and mathematics
(STEM) workforce, and to what extent would failure to diversify our
workforce hinder our national and economic security?
Answer. Studies have shown that diversity fosters greater outcomes
in problem solving than monoculture groups. Diversity increases problem
solving skills, creativity and conflict resolution. Black and Hispanic
workers continue to be underrepresented in the STEM workforce, while
White and Asian workers are overrepresented. STEM degrees lead to high-
impact career opportunities, which can lift up communities.
Investment in our Nation's STEM education system is essential to
maintain and grow our economy. STEM graduates drive innovation in
health, clean energy, technology, aerospace and infrastructure, but our
states' industries often turn elsewhere for workers because our
universities cannot keep up with demand due to the lack of faculty,
space and resources. Insufficient STEM capacity denies our citizens
access to high-impact and high-earning careers. This inaccessibility is
particularly acute for individuals traditionally underrepresented in
these fields.
A thriving economy demands a well-rounded workforce for long-term
competitiveness. Failing to prepare our citizens for the work of the
future compromises our country's long-term competitiveness, security
and economic stability. To remain competitive on a global stage we must
increase the diversity in STEM and leverage all of our talent.
Question 2. What role do historically black colleges and
universities (HBCUs) and other minority-serving institutions play in
helping to build a diverse workforce? What lessons can universities and
other institutions draw from the success of HBCUs in supporting our
STEM workforce?
Answer. HBCUs play an important role in building a more diverse
workforce. A quarter of Black STEM graduates come from HBCUs. Forty-six
percent of Black women who earned STEM degrees between 1995-2004
graduated from HBCUs. The institutions of origin for 30 percent of
Black graduates earning doctorates in science and engineering were
HBCUs.
There is a greater need for equity throughout our educational
system and in industry hiring. HBCUs serve students who are first-
generation, low-income, and welcome people from any race to attend.
Universities across the Nation would be well served by mirroring this
inclusive approach and support diverse STEM students, staff and faculty
throughout their educational and professional journeys. There are
numerous opportunities for government, industry and academia to come
together to create STEM career pathways for all of our citizens.
Rural Institutions: Georgia is a proud agricultural state, and I am
particularly excited for how advanced technologies will help improve
precision agriculture and otherwise support America's farmers. I am
concerned, however, by the underrepresentation of rural students and
rural colleges and universities among America's leading research
institutions and the effect this will have in fields such as precision
agriculture.
Question 3. What barriers do rural students face in entering the
workforce in advanced technologies such as artificial intelligence and
quantum computing? How can Congress help address these barriers?
Answer. Rural students face numerous barriers in accessing
education that would allow them to pursue careers in advanced
technologies. Rural schools are often disadvantaged by the way
education funds are calculated and distributed as smaller enrollments
result in fewer dollars. In addition, child poverty levels in rural
areas can be double that of urban communities. Rural residents attend
college at a lower rate than urban residents (19 percent to 35
percent).
The lack of financial support from the community and lack of
visibility with private funders challenges rural schools. Urban schools
often receive priority with grants and research to improve their
conditions while rural schools garner less attention. Teacher pay is
often much lower in rural areas than in urban areas and that impacts
recruitment and retention of new and qualified teachers. Rural parental
and community opinions of STEM education are often incongruent with the
need for students to learn certain skills and principles of the field
to be competitive.
Some problems exist for rural schools to take advantage of
supplemental government grants and funding, so Congress should be
mindful of these potential pitfalls. For example, the Rural Education
Achievement Program (REAP) designed to assist rural schools with
administrative challenges in their school systems. Stipulations on fund
use sometimes create unintended burdens (e.g., requiring that a
percentage of the award be set aside for supplemental educational
services, but this is a costly hardship for remotely located schools).
Question 4. What barriers do rural colleges and universities and
their faculties face in accessing and supporting research related
advanced technologies such as artificial intelligence and quantum
computing? How can Congress help address these barriers?
Global Collaboration and Competition: I have long believed that
America must invest in innovation and advanced technologies to maintain
our global standing. I am deeply alarmed that our global competitors,
such as the Chinese Communist Party, have heavily invested in
technology research and used it to oppress its own people and bolster
autocratic regimes throughout the world. To keep America on the leading
edge, Congress must strike the appropriate level of controls on
innovation and intellectual property to protect our national and
economic security without overly hindering scientific innovation and
international collaboration.
Answer. In a competitive global market, there is great demand for
qualified individuals in STEM fields. Rural schools struggle to prepare
people to compete at an early age in STEM fields. To maintain our
economic and intellectual global leadership we must offer equal STEM
educational opportunities to all students. Rural communities lack
access to resources for STEM education, incongruent local values
between local culture and economic demand, rural poverty, and gaps in
outreach to rural communities for aid and research.
Jobs in AI are well suited for remote work in rural areas if tools
are accessible. AI could help address problems in rural communities
such as the opioid crisis and crop production in farming and
agriculture. STEM education is the foundation for lucrative careers in
AI and data science. Expanding digital connectivity and investing in
workforce development and government, academia and industry
partnerships could transform local economies and reduce disparities in
earning capacities.
References: Maskey, S. (2021, April 13). We need to teach AI in
rural America. Forbes. https://www.forbes.com/sites/forbestechcouncil/
2021/04/13/we-need-to-teach-ai-in-rural-america/?sh=2352852e418b
Harris, R.S. and Hodges, C.B. (2018). STEM education in rural
schools: Implications of untapped potential. National Youth-At-Risk
Journal, 3(1). https://files.eric.ed.gov/fulltext/EJ1269639.pdf
Question 5. What are the key concerns that Congress should keep in
mind as we work to achieve this balance? Please share any examples of
Federal programs that you believe do not strike the right balance.
Answer. Rural America is facing a maelstrom of disparities,
exacerbated by the widening technology divide. Many rural areas lack
reliable cost-effective energy resources. Deficient digital
connectivity in rural areas has led to poor health and economic under-
development. We must re-think digital innovation as a process to
reconnect and remap the economies in rural and urban areas.
Institutions of higher education are uniquely positioned to take
advantage of the powerful tools developed in the digital industry and
lead a new type of digital ecosystem due to their strengths in
fostering entrepreneurship, workforce education, collaboration with the
private sectors and national laboratories, promoting diversity and
inclusion and being a natural nexus to technology and society.
I urge you to continue to invest in our Federal science agencies
and initiatives empowered by CHIPS. Sustained Federal investment in
these programs are essential for our Nation to remain a leader in a
fierce global landscape, to leverage opportunities for collaboration
between government, academic, and business sectors, and to build a
workforce that reflects the rich diversity of our Nation.
______
Response to Written Questions Submitted by Hon. Maria Cantwell to
William B. (Trey) Breckenridge III
Supercomputing Leadership: You noted in your testimony that the
United States is losing its lead in high-performance computing to
China. In 2015, the United States had nearly twice as many of the Top
500 supercomputers as China. Today, however, China leads with 173
supercomputers, compared to 123 in the United States. The U.S.
Department of Energy (DOE) owns four of the top ten fastest
supercomputers in the world according to the June 2022 Top500 list. The
Frontier supercomputer at the DOE's Oak Ridge National Laboratory
(ORNL) is ranked as the fastest computer in the world.\3\
Question 1. Given that other computational capabilities are readily
available to a broad group of industry, academic, and Government
entities, what importance should policymakers assign to ensuring that
the United States has access to the world's fastest supercomputers? Are
there specific examples where the United States lacking these
capabilities has led--or will lead--to a strategic disadvantage?
Answer. Supercomputing access is the future of R&D in all STEM
fields. The development of new technologies and modeling capabilities
simply cannot happen without high performance computing capability.
While access to supercomputing resources has significantly increased
over the last decade or so, there is not adequate capacity in those
systems to meet the needs of those communities utilizing them. As was
the case when we began improving Internet connectivity, those who are
limited in high performance computing capacity simply will be left
behind globally to those who have the robust, high-speed tools for data
analytics, modeling capacity, and research throughput.
However, simply increasing HPC capacity is not sufficient to ensure
high-end computing needs are met. While the number of systems on the
TOP500 list is an indicator of availability of access to HPC, it is not
an indicator of overall capability. For instance, the Frontier system
at Oak Ridge National Laboratory has more computing capacity than the
remaining 127 U.S. systems of the TOP500 list added together. To
address the massive, highly complex workloads of grand challenge
problems, such as quantum mechanics, climate change, and security, and
to remain competitive globally in these critical areas, the U.S. must
fund more capability-class HPC systems--systems with significantly
higher performance than the lower 90 percent of the TOP500--and make
these systems available to U.S. scientists and researchers.
Compute Access: Academic institutions represented the majority of
large-scale AI experiments from the 1960s until 2010. However, over the
past decade the industry-academic balance has altered, with the vast
majority of large-scale AI research being carried out by industry
rather than academia. Private industry is leading in AI development,
because they have the large-scale resources, and therefore the access,
to carry out frontier AI research. Building large-scale AI models can
cost tens of millions of dollars and require 300,000x more computing
power than experiments conducted in the prior decade.
Question 2. What are the specific gaps you are seeing in the
availability of AI computing resources in industry, government, and
academia, and to what extent do gaps differ in the regulated (e.g.,
classified) and unregulated spaces? Besides the National AI Research
Resource, what else can the U.S. government do to catch up to private
industry, or better collaborate with private industry, in AI
development?
Answer. Industry has indeed made major investments in R&D related
to AI technologies. Funding programs should be designed that encourage
more government-industry-academic partnerships. Rather than viewing
them as competitors, collaboration will be critical if we are to move
forward as a Nation in leading AI technology development and
application.
______
Response to Written Questions Submitted by Hon. Kyrsten Sinema to
William B. (Trey) Breckenridge III
Drought and Fire Monitoring at NOAA: Fire season began in late
April in Arizona, as the Southwest experiences the most severe drought
in twelve hundred years. The Tunnel Fire north of Flagstaff burned over
twenty thousand acres while the Crooks Fire consumed over six thousand
acres south of Prescott. After the fires were extinguished, burn scars
have led to significant flooding issues in Flagstaff and other areas in
northern Arizona.
Arizonans rely on National Weather Service predictions and National
Oceanic and Atmospheric Administration (NOAA) data to predict fire
trajectory and to determine impacts to landscapes and human health. It
is important that NOAA and the Weather Service have the resources and
expertise needed to predict weather and drought conditions.
Earlier this year, NOAA launched two new supercomputers, including
the ``Cactus'' supercomputer in Phoenix, Arizona, which operates at a
speed of 12.1 petaflops.
Question 1. The bipartisan infrastructure law included $80 million
dollars for NOAA high-performance computing technology and $50 million
for wildfire prediction, detection, and forecasting. How will these
resources allow NOAA to improve drought and wildfire prediction
forecasting?
Answer. High performance computing is critical to improve wildfire
and drought forecasting. As modeling capabilities improve, the need for
computational speed goes up exponentially. Our capacity to generate
data to improve models and the accuracy thereof can be greatly limited
if research does not have the computational capacity to develop and
adequately use the data generated. In the same way, operational models
require near real-time processing if predictions are to be relevant.
STEM Education in CHIPS and Science Act: The bipartisan CHIPS and
Science Act authorized $13 billion over five years for the National
Science Foundation (NSF) to put towards Science, Technology,
Engineering, and Mathematics (STEM) education. The legislation directs
the NSF to explore opportunities to engage students from groups
historically underrepresented in STEM fields as well as rural areas to
expand the pipeline to help ensure the United States has the STEM
workforce we need for the Twenty-First Century.
Question 2. How will the CHIPS and Science legislation expand STEM
education and, in turn, America's STEM workforce? Why is it important
for Congress to fund the investments in STEM education made in the
legislation?
Answer. The CHIPS and Science legislation provides an opportunity
for substantial increase in NSF funding at critical juncture. Other
countries have made major advances in STEM areas, and if the U.S. is to
regain its leadership role investments like this are crucial.
Importantly, funds are set aside in this legislation to ensure that
more institutions in more states are able to participate. This will
allow many, many more students to work in NSF-funded projects,
broadening participation and recognizing that workforce development in
STEM fields is critical throughout the Nation, and not just in a few
states.
Semiconductor Investments: Once the global leader, America now only
accounts for approximately 12 percent of the world's semiconductor
manufacturing. Recognizing this important national security issue, I
worked with my colleagues to ensure the CHIPS and Science Act also
appropriated $52 billion for the Commerce Department to establish
grants to jump-start investments in America's domestic semiconductor
industry. These investments will support tens of thousands of Arizona
jobs at facilities managed by Intel, TSMC, and other Arizona
semiconductor companies.
Question 3. How does the CHIPS and Science Act and the law's
semiconductor investments illustrate the United States' commitment to
being a global leader in this field and what further actions can be
done to expand U.S. leadership in this field?
Answer. The funding commitments that the CHIPS and Science
legislation makes is a bold statement globally about the U.S.
commitment to leadership in STEM fields. In addition, the funding
recognizes the critical national security needs in onshore chip
manufacturing. Investments in research and ensuring the environment for
onshore manufacturing for the development of new technologies will be
critical to ensure that the U.S. adequately addresses these national
security needs. In particular, additional actions should be taken to
ensure companies have access to capital for the manufacturing
investments necessary as well as a trained workforce capable of meeting
the demands from this added manufacturing capacity.
______
Response to Written Questions Submitted by Hon. Raphael Warnock to
William B. (Trey) Breckenridge III
Workforce Diversity:
Question 1. What is the importance of fostering greater diversity
within America's science, technology, engineering, and mathematics
(STEM) workforce, and to what extent would failure to diversify our
workforce hinder our national and economic security?
Answer. Our Nation is founded on the principle of freedom and
opportunity for all. We cannot afford to see a lack of diversity in
STEM fields if we are to achieve our potential. We as a country cannot
be successful if we in fact leave a segment of our population behind in
technology development.
Question 2. What role do historically black colleges and
universities (HBCUs) and other minority-serving institutions play in
helping to build a diverse workforce? What lessons can universities and
other institutions draw from the success of HBCUs in supporting our
STEM workforce?
Answer. HBCUs and MSIs have historically played a vital role in
ensuring higher education opportunities for all citizens of the US. As
STEM fields evolve, collaboration between all institutions will be even
more important than in the past. HBCUs and MSIs, in particular
partnering with other research universities, will have exceptional
opportunities to ensure that research funding and workforce development
is broad across all demographics within the U.S.
Rural Institutions: Georgia is a proud agricultural state, and I am
particularly excited for how advanced technologies will help improve
precision agriculture and otherwise support America's farmers. I am
concerned, however, by the underrepresentation of rural students and
rural colleges and universities among America's leading research
institutions and the effect this will have in fields such as precision
agriculture.
Question 3. What barriers do rural students face in entering the
workforce in advanced technologies such as artificial intelligence and
quantum computing? How can Congress help address these barriers?
Answer. So much of the research funding in recent years has gone to
a few institutions in a few states, thereby limiting access to
undergraduate and graduate research opportunities across the Nation.
The CHIPS and Science legislation has provided important safeguards to
ensure that funding is provided more broadly to rural and
underrepresented populations. This legislation ensures that quality
will not be lessened, but many more students will have access to these
resources so that we as a Nation can grow across the entire country.
Question 4. What barriers do rural colleges and universities and
their faculties face in accessing and supporting research related
advanced technologies such as artificial intelligence and quantum
computing? How can Congress help address these barriers?
Answer. Access to funding has been difficult for a variety of
reasons, including perceptions of quality, lack of research
infrastructure, lack of state investment, and higher demands on faculty
at institutions with less funding. In particular, a need exists to
invest more resources in equipment and research infrastructure if rural
colleges and universities are to be successful in R&D activities.
Global Collaboration and Competition: I have long believed that
America must invest in innovation and advanced technologies to maintain
our global standing. I am deeply alarmed that our global competitors,
such as the Chinese Communist Party, have heavily invested in
technology research and used it to oppress its own people and bolster
autocratic regimes throughout the world. To keep America on the leading
edge, Congress must strike the appropriate level of controls on
innovation and intellectual property to protect our national and
economic security without overly hindering scientific innovation and
international collaboration.
Question 5. What are the key concerns that Congress should keep in
mind as we work to achieve this balance? Please share any examples of
Federal programs that you believe do not strike the right balance.
Answer. Investments in research within the U.S. is a critical step,
and Congress is to be applauded for the CHIPS and Science legislation
that will be a major step forward in ensuring national competitiveness.
I believe Congress has done a very good job in striking the right
balance between controlling and stimulating innovation. While recent
policies restricting access to R&D to foreign entities have been
difficult to implement, I recognize the critical need for these steps.
However, funding to cover the added costs of these controls and
restrictions are born by individual universities, and additional
Federal funds are critically needed if rural universities are to remain
competitive.
______
Response to Written Questions Submitted by Hon. Maria Cantwell to
Dr. Bob Sutor
QIS Technical Challenges: CHIPS and Science has authorized National
Institute of Standards and Technology funding for research and
development of quantum cryptography and post-quantum classical
cryptography standards, as well as quantum networking, communications,
and sensing technologies. By studying these fields, we can understand
the impacts of quantum information sciences on cybersecurity, secure
communications, and medical devices.
Question 1. What are the main technical challenges facing the
quantum information science field, and how can the United States
Government help meet those challenges?
Answer. The primary challenge for quantum computing systems is
building large and powerful enough systems. We often measure ``large
enough'' in terms of the number of qubits, but this is not a good
metric. Having many error-prone qubits is not necessarily better than
having fewer qubits with higher fidelity. The quantum programming model
requires that qubits be connected to create entanglement. We will
likely need hundreds of thousands of qubits arranged in arrays of
interconnected quantum cores that can operate at very low error rates
for application to the promised use cases in AI, optimization, and
simulation of physical systems. We need breakthroughs in physics,
engineering, and manufacturing to get there.
Another challenge facing quantum computing is that its programming
is not a simple variation or extension of how we code software for
classical applications. The model is entirely different and requires
new training materials and courses. In my book, Dancing with Python, I
show how this integrated approach can teach students, even at the high
school level, to start understanding and coding quantum algorithms. We
need national STEM programs to make quantum software development
pervasive throughout our computer science curricula.
We can also make physics education more accessible via the cloud.
Quantum sensors work by detecting subtle effects and variations in
atomic configurations. This allows us to determine changes in speed,
rotation in three dimensions, and gravity. With this technology, we can
also build high-resolution atomic clocks and Radio Frequency receivers
for consumer, military, and intelligence use. We need to train more
students at the university level to move from the theoretical aspects
of quantum computing to applied quantum engineering. Quantum matter
machines, such as the quantum emulator named Albert from ColdQuanta,
can give students a hands-on cloud-based interactive platform for
experimenting with the physical processes at the core of quantum
sensors. Just as quantum computing was jump-started in 2016 when the
first system was put on the cloud, we can do the same to accelerate
skills development for quantum sensors and other devices.
A major issue facing America's quantum industry is supply chain.
Many essential enabling technologies for quantum technology are
available only from a handful of vendors, many of which are overseas.
For example, ColdQuanta's cold atom approach to quantum computing uses
many different laser systems to trap, cool, and manipulate atoms. These
are very specialized lasers, with only a few vendors offering systems
with the necessary requirements. Sometimes we have to build our own
laser systems in-house to meet our needs. Another key enabling
technology that will allow quantum technology to become much more
compact is Photonic Integrated Circuits or PICs. This technology is to
light what microchips are to electronics; with a PIC, you can build an
entire photonic system on a chip.
We must support a robust domestic supply chain. We cannot afford to
allow quantum and quantum-enabling manufacturing to go overseas. To
avoid that, we need to invest now in small, domestic manufacturers,
building that supply chain and providing high-paying jobs for
Americans.
Overcoming Valley of Death: You mentioned in your testimony that
the U.S. Government's investment in quantum startups is scattered,
without support for integration into deployed systems of record. The
CHIPS and Science Act specifically aimed to help accelerate startup
development, including through entrepreneurial fellowships, testbeds,
and demonstration projects within the National Science Foundation
Technology and Innovation Partnerships directorate.
Question 2. What specific policy recommendations do you have to
improve U.S. investment in startups to help them cross the ``valley of
death''? From your perspective, do quantum and AI technologies face
unique challenges, such as greater capital investment that ought to be
addressed?
Answer. The ``valley of death'' concerning quantum startups and
manufacturers is the gap between prototyping/low-volume production and
high-volume production. Hardware typically requires greater and longer-
term investment than software because of the prototype, development,
and product cycle, together with supply chain fulfillment issues and
delays. In addition, it will be several years, perhaps five to ten,
until we have powerful enough quantum computers for our intended
applications. The venture capital and general investment community may
not have the patience to wait that long for their ROI.
Statistically, 90 percent of all startup companies in the U.S.
fail. It is essential that promising small and medium-sized companies
within the quantum ecosystem be supported until they can reach
manufacturing scale and profitability for this technology that will be
so important for our country's economic success and security.
Some of this support can come in the form described in our response
to your Question 3: departments and agencies should not penalize
contractors for accepting cost and schedule risk for the inclusion of
highly promising quantum solutions, given the strategic importance to
the Nation of accelerating the development and maturation of this high-
payoff technology.
It is also critical that any legislation maintains quantum
technology neutrality. At this exciting stage of technological
development, there is a diversity of quantum modalities. To make
quantum systems, people use atoms, ions, superconducting systems,
nitrogen vacancies, photons, and more. While we at ColdQuanta believe
that the neutral atom approach is the surest bet, we, as an industry,
are still determining which modality will deliver the first fully
functional and error-corrected quantum computer. In the case of quantum
computing, if we need hundreds of thousands of qubits for practical
quantum advantage, we are years away from declaring winners and losers.
The best technology today only has dozens or a few hundred qubits. If
the United States government were to pick a ``winner'' quantum
modality, it would be a significant gamble and risk crippling promising
technologies that are not yet fully developed. Supporting all
modalities equally is the best way to ensure a robust and productive
quantum ecosystem where development can be accelerated.
National Quantum Initiative:
Question 3. What goals should the Committee consider in
reauthorizing the National Quantum Initiative?
Answer. Congress and the Administration have taken significant and
very wise initial steps to recognize the power and potential of quantum
science and engineering advancement and the out-sized role such
progress will have for all aspects of our society. The bipartisan
National Quantum Initiative Act of 2018 was an excellent step requiring
a government plan to accelerate quantum science and engineering
development in a coordinated and strategic fashion.
Four areas that would be valuable for the Committee to consider are
workforce development, manufacturing support, more centralized
organization of quantum programs within the government, and streamlined
acquisition processes that encourage quantum technological development.
The quantum industry is growing rapidly, and its workforce needs to
grow with it. To encourage students to pursue careers in the quantum
industry, we need to make quantum education more accessible. Offering
funding to high schools and community colleges to develop quantum
educational programs, particularly with affordable access to cloud-
based quantum computers and quantum matter machines, will ensure that
students get early exposure to quantum science. Students studying
quantum-related fields could be offered scholarships, and workers who
spend several years within the quantum industry, particularly for
technician-level jobs, could be offered loan forgiveness to encourage
their participation in the quantum workforce.
In the interest of economic growth and national security, we urge
the Committee to take steps to support a robust domestic supply chain.
By investing in small, domestic manufacturers, we can build that secure
supply chain and provide high-paying jobs for Americans.
We also urge the Committee to continue its leadership role by
encouraging key government departments and agencies to move more
quickly to tailor internal organizational practices in recognition of
the critical strategic importance of quantum development. Departments
and agencies must move swiftly and decisively to adopt detailed program
planning, budgeting, and program acquisition policies/procedures to
accelerate quantum-based technology advancement. This includes creating
a centralized system to plan, track, monitor, assess, and ensure high
payoff developments are sufficiently resourced for all viable quantum
science R&D projects across the Department/agency enterprise. It also
entails the establishment of effective mechanisms for Federal agencies
to share/coordinate quantum R&D activity plans and results to maximize
the speed and quality of quantum development.
We urge the Committee also to consider legislation to encourage
departments and agencies to modify their program acquisition procedures
expressly to allow for the insertion of viable quantum technology
solutions. Since we are early in the quantum technology development
cycle, prime contractors should not be penalized or discouraged for
proposing to accept higher risk/higher reward quantum technology
solutions in the acquisition selection and contract negotiation
processes. Departments and agencies should not penalize contractors for
accepting cost and schedule risk for the inclusion of highly promising
quantum solutions, given the strategic importance to the Nation of
accelerating the development and maturation of this high-payoff
technology.
______
Response to Written Questions Submitted by Hon. Kyrsten Sinema to
Dr. Bob Sutor
STEM Education in CHIPS and Science Act: The bipartisan CHIPS and
Science Act authorized $13 billion over five years for the National
Science Foundation (NSF) to put towards Science, Technology,
Engineering, and Mathematics (STEM) education. The legislation directs
the NSF to explore opportunities to engage students from groups
historically underrepresented in STEM fields as well as rural areas to
expand the pipeline to help ensure the United States has the STEM
workforce we need for the Twenty-First Century.
Question 1. How will the CHIPS and Science legislation expand STEM
education and, in turn, America's STEM workforce? Why is it important
for Congress to fund the investments in STEM education made in the
legislation?
Answer. The provisions in the CHIPS and Science legislation will go
a long way toward strengthening and diversifying America's STEM
education and workforce. These provisions will require Federal agencies
and universities to identify and lower barriers to the recruitment,
retention, and advancement of women, minorities, and other
underrepresented groups in STEM. They will also require much-needed
data collection on the demographics of the STEM workforce, provide
insights on reasons for poor recruitment and retention, and inform best
practices for making the STEM workforce inclusive. Funding these
provisions will yield valuable gains in the development of our STEM
workforce and accelerate its diversification.
We agree with the provisions and requirements in the Chips and
Science Act--Title V: Broadening Participation in Science document at
https://science.house.gov/imo/media/doc/STEM
percent20Participation.pdf.
Semiconductor Investments: Once the global leader, America now only
accounts for approximately 12 percent of the world's semiconductor
manufacturing. Recognizing this important national security issue, I
worked with my colleagues to ensure the CHIPS and Science Act also
appropriated $52 billion for the Commerce Department to establish
grants to jump-start investments in America's domestic semiconductor
industry. These investments will support tens of thousands of Arizona
jobs at facilities managed by Intel, TSMC, and other Arizona
semiconductor companies.
Question 2. How does the CHIPS and Science Act and the law's
semiconductor investments illustrate the United States' commitment to
being a global leader in this field and what further actions can be
done to expand U.S. leadership in this field?
Answer. The CHIPS and Science Act takes bold and critical steps
towards re-establishing a robust domestic semiconductor supply chain.
Recent chip shortages have affected consumer goods, from cars to
computers, mainly due to reliance on a highly homogenous, overseas
supply chain. Investing in our domestic capacity to produce
semiconductors will strengthen our national security, limit the
economic distress that can result from supply chain problems, and, as
you aptly note, create many high-paying tech jobs for Americans. But we
should not stop there.
Many quantum enabling technologies are only available from a few
small vendors, and many of those vendors are overseas. For our national
security and economic growth, we must develop a robust domestic quantum
supply chain by investing in small domestic manufacturers.
It is important to note that many quantum technologies are not
packaged as ``chips.'' While semiconductors are used in the control
systems for cold atoms, ion traps, and photonic quantum devices, the
computer or sensor may be within a glass cell, with lasers performing
many functions.
______
Response to Written Questions Submitted by Hon. Ray Ben Lujan to
Dr. Bob Sutor
Development of Quantum Sensors: Dr. Sutor, you noted in your
testimony that sensors are a near-term application of quantum
technologies. You cite inertial sensors and clocks as particularly
useful implementations that soon could be fielded and commercialized.
Our National Laboratories are leaders in the field of quantum
engineering. For example, Sandia's Quantum Information Program has
improved our understanding and mastery of quantum systems, enabling
cutting edge sensor systems. Congress dedicated significant resources
to the expansion of quantum technologies in the CHIPS and Science Act.
Question 1. Yes or No, should we be leveraging our National Lab's
assets and expertise to rapidly mature and commercialize these emerging
technologies?
Answer. Yes.
Question 2. How can we better support public-private partnerships
to accelerate the development of promising sensing applications?
Answer. We urge the Committee to consider legislation to encourage
departments and agencies to modify their program acquisition procedures
expressly to allow for the insertion of viable quantum technology
solutions. Since we are early in the quantum technology development
cycle, prime contractors should not be penalized or discouraged for
proposing to accept higher risk/higher reward quantum technology
solutions in the acquisition selection and contract negotiation
processes. Departments and agencies should not penalize contractors for
accepting cost and schedule risk for the inclusion of highly promising
quantum solutions, given the strategic importance to the Nation of
accelerating the development and maturation of this high-payoff
technology.
We should specifically look at how the public-private partnerships
can accelerate the time from prototype to system-of-record for higher-
precision atomic clocks, inertial sensors, and Radio Frequency
receivers. It's not enough to develop the individual components and
sub-components of quantum computers, quantum sensors, and quantum
communication equipment. The partnerships should aim to create complete
solutions that can be quickly commercialized.
The most critical gap in the development of quantum sensing
applications is the scalable manufacture and assembly of these complete
systems. The manufacture of the core photonic and quantum components
has been demonstrated, and the assembly technologies are available. An
ideal, shovel-ready public-private partnership is to build out these
demonstrated capabilities as a flexible, for-profit service. The
service will support the critical manufacturing needs of the quantum
sensor industry as it develops while producing the core assemblies for
commercial quantum products.
Question 3. Where do you anticipate that quantum sensors may be
most useful? You cite GPS systems as a vulnerability that could be
mitigated using quantum sensors. Would these efforts make our Nation
more secure?
Answer. Quantum technology will give us next-generation navigation
with the highest possible precision. There is no scale of resolution
finer than quantum. Our current GPS has several weaknesses. GPS signals
can be affected by weather and other atmospheric conditions. More
significant problems, especially for those concerned with security, are
``GPS jamming'' or ``denial,'' where a signal is turned into noise, or
``GPS spoofing,'' where a valid signal is replaced by a stronger and
nefariously incorrect one. You don't want to be on a plane that thinks
it is hundreds of miles away from its actual location. This practice is
known and common in war zones, and denial or spoofing of GPS in a major
city could snarl much of its transportation with apparent implications
for safety and commerce.
Financial transactions and cellular and power infrastructure are
dependent on GPS. Financial transactions across networks use GPS for
timestamping. Accuracy is essential in high-speed applications to know
the exact sequence of transactions. Cellular base stations can use GPS
to synchronize their times to use the broadband spectrum more
precisely. Power networks are complicated these days, with multiple
energy sources and, often, bidirectional flow. Time synchronization
from GPS is used in some grids to optimize and balance electricity
distribution.
Quantum atomic clocks combined with quantum inertial sensors will
be able to provide what is, in effect, onboard dead reckoning. There
will be no need to connect to satellites or external sources that can
be blocked or manipulated. We will likely migrate to this kind of
quantum-based solution for PNT--Positioning, Navigation, and Timing.
This technology will be useful to the warfighter and make our Nation
more secure; the military and defense will most likely be the first
users of these solutions. But, like GPS, businesses will be able to
develop commercial products, and we could use them in our everyday
lives, such as in fully autonomous self-driving cars that cannot lose
signal in tunnels or bad weather.
Quantum atomic clocks are already used today in GPS satellites, and
they will eventually become pervasive elsewhere as they become smaller
and less expensive. They will appear in our network, cloud data
centers, cell phone towers, ATMs, planes, and ships, as well as in our
cars and phones. Not only will they operate independently or in
ensembles and be highly accurate, but they will also maintain that
accuracy for weeks or months before resynchronizing.
Beyond PNT, quantum sensors will have many uses that will make our
Nation more secure. For example, quantum Radio Frequency (RF) sensors
are significantly more sensitive than traditional ones and will enable
warfighters to detect RF signals that are undetectable today. We can
even measure gravity fluctuations with quantum gravimeters. These can
determine changes in the earth's density and help discover new
resources. Other applications include safety and recovery operations,
such as finding voids in building collapses after an earthquake, or,
more salient to national security, discovering underground structures
or tunnels that adversaries may have constructed out of sight. We might
even get early alerts for natural disasters such as landslides and
sinkholes.
______
Response to Written Questions Submitted by Hon. Raphael Warnock to
Dr. Bob Sutor
Workforce Diversity:
Question 1. What is the importance of fostering greater diversity
within America's science, technology, engineering, and mathematics
(STEM) workforce, and to what extent would failure to diversify our
workforce hinder our national and economic security?
Answer. The workforce needs of the quantum industry are growing as
rapidly as the industry is. To meet these needs, we cannot afford to
overlook people of talent who have been historically underrepresented
in quantum physics, particularly women and people of color. Embracing
all forms of diversity as we grow the quantum workforce will not only
maximize the pool of talent but having a diverse workforce is highly
beneficial in the workplace in terms of productivity, creativity, and
problem-solving. ColdQuanta is dedicated to building a diverse quantum
workforce with internal education efforts to ensure we provide a highly
inclusive work environment and external efforts aimed toward equitable
recruitment and hiring practices. Diversity is an important value that
will make our domestic quantum workforce stronger and more competitive
globally.
The United States needs the best talent to compete economically and
ensure that we always have the most advanced technology for our
national security. The coming quantum era will require a broad range of
skills, talents, and backgrounds to secure our success. We commonly
hear about academic and industrial research results from those with
doctorates. For quantum to become pervasive and practical, we need
hardware and software engineers, technicians, and manufacturing
workers. Quantum and classical computing will co-exist. While some
people working in traditional computing will develop quantum skills, we
should primarily consider the new talent we need for the very different
aspects of quantum. Our needs will only add to whatever shortcomings we
have now. Those needs are diverse, as must be the people who fulfill
them.
Question 2. What role do historically black colleges and
universities (HBCUs) and other minority-serving institutions play in
helping to build a diverse workforce? What lessons can universities and
other institutions draw from the success of HBCUs in supporting our
STEM workforce?
Answer. According to a 2019 paper in the Educational Researcher
journal,\1\ ``STEM is the only field where Black and Latina/o youth are
significantly more likely than their White peers to switch and earn a
degree in another field.'' Since quantum technology is so different
from its classical counterpart, now is the time to remedy the STEM
disparity regarding diversity and inclusion. We have a new opportunity
to ``do it right.'' We can build a diverse workforce in a novel
industry, close the STEM participation gaps for quantum, and keep
people in the field. Quantum is not just physics: its study includes
mathematics, computer science, chemistry, and engineering, and it has
applications in AI, materials science, logistics, finance, and any
field that uses optimization techniques. HBCUs can teach quantum as a
multidisciplinary field that better reflects the broad range of jobs
that will be available in the industry.
---------------------------------------------------------------------------
\1\ ``Does STEM Stand Out? Examining Racial/Ethnic Gaps in
Persistence Across Postsecondary Fields.'' https://
journals.sagepub.com/doi/full/10.3102/0013189X19831006?journalCode=edra
---------------------------------------------------------------------------
Providing a high standard of quantum education does not need to be
burdensome in terms of time or resources for the faculty offering it.
Students, professors, and researchers can access quantum computers and
emulators via the cloud. There is no need to install a quantum system
locally. For example, ColdQuanta will offer quantum matter machines and
quantum computers on the cloud. With them, users will be able to create
ultra-cold quantum matter or perform quantum calculations remotely.
Further, much of the software for using quantum computers is open
source, such as Cirq and Qiskit. The essential resources for learning
about and using quantum computers are available from our rural areas to
our biggest cities.
Rural Institutions: Georgia is a proud agricultural state, and I am
particularly excited for how advanced technologies will help improve
precision agriculture and otherwise support America's farmers. I am
concerned, however, by the underrepresentation of rural students and
rural colleges and universities among America's leading research
institutions and the effect this will have in fields such as precision
agriculture.
Question 3. What barriers do rural students face in entering the
workforce in advanced technologies such as artificial intelligence and
quantum computing? How can Congress help address these barriers?
Answer. Access is a significant barrier to entry for the rural
student. In the past, to do quantum science or receive quantum
education, you had to be physically present in a laboratory, but not
anymore. The remote availability of quantum computers and emulators via
the cloud makes quantum science much more accessible. Open-source
software and online courses make it easier for the rural student to
learn quantum science and for faculty at rural institutions to offer
such education.
Internships are crucial to getting practical experience and
demonstrating proficiency in STEM areas like AI and quantum. During the
pandemic, many internship programs were virtual, and students could
participate from anywhere. Now, fewer remote internships are available,
and there is often an assumption that face-to-face is required. In some
cases, such as hardware, one must be where the facilities are. For
software, while ``back to the office'' may be advantageous, it has
disadvantages for those who cannot travel.
Congress can help address the barriers for rural students by:
ensuring equitable cloud access to quantum computers and
emulators for students, faculty, and researchers,
ensuring that U.S. government labs and agencies offer
virtual options for internships whenever possible,
fully reimbursing students for all reasonable travel and
living expenses when temporary relocation for an internship is
necessary,
offering scholarships for students who choose to pursue
quantum education and loan forgiveness for students who go to
work in the quantum industry; and
encouraging commercial organizations to consider students
from a broad range of geographic locations.
While K-12 curricula are usually in the domain of the states,
education on quantum technologies and data science should extend
downward into high schools. National educational standards for the
technologies most important to our economic and security interests
should be considered. High schools in rural areas should be supported
by training educators and giving students low-cost access to learning
materials and cloud-based technology.
Question 4. What barriers do rural colleges and universities and
their faculties face in accessing and supporting research-related
advanced technologies such as artificial intelligence and quantum
computing? How can Congress help address these barriers?
Answer. A significant barrier for rural institutions and faculties
is cost. To become proficient in and help develop leading-edge
technology, one must have access to leading-edge technology. State-of-
the-art research and graduate work in physics and engineering require
investment in the equipment necessary. However, establishing and fully
equipping a quantum science laboratory is time-consuming and expensive.
With cloud-based solutions, this is not always necessary, and many of
these costs can be avoided and time saved. Congress must ensure that
access to cloud-based AI and quantum computing software is available
and funded for research and education at rural institutions. Congress
can also direct agencies like the National Science Foundation to fund
quantum projects and equipment and develop quantum programming at rural
schools.
Another possibility is to institute a more aggressive ``rotating
scholar'' program where professors and researchers from less rural and
academically higher-ranked institutions visit the rural colleges and
universities for at least one term. These visiting positions exist
today but could be increased five to tenfold to accelerate AI, quantum
research, and education at rural institutions.
Global Collaboration and Competition: I have long believed that
America must invest in innovation and advanced technologies to maintain
our global standing. I am deeply alarmed that our global competitors,
such as the Chinese Communist Party, have heavily invested in
technology research and used it to oppress its own people and bolster
autocratic regimes throughout the world. To keep America on the leading
edge, Congress must strike the appropriate level of controls on
innovation and intellectual property to protect our national and
economic security without overly hindering scientific innovation and
international collaboration.
Question 5. What are the key concerns that Congress should keep in
mind as we work to achieve this balance? Please share any examples of
Federal programs that you believe do not strike the right balance.
Answer. I agree that striking the right balance is critical. At one
extreme, some could say, ``control everything that might pose a threat
anywhere in its future development.'' At the other extreme, some might
want to sell as much as possible to anyone who wants a technology to
reap the economic benefits. But neither extreme makes sense, and you
are correct that as we develop export controls as a nation, we need to
find the sweet spot or the natural balance point.
In the case of quantum computing technology, we can ask, ``how
large and powerful a quantum computing machine can we sell
internationally without endangering our national security?''. It's not
an easy question to answer. Academics and industrial researchers are
starting to build quantum computers from various core qubit
technologies, including cold atoms, ion traps, superconducting, and
photonics. There are seven or eight characteristics that play off
against each other and make it difficult to know which approach will
ultimately be best to scale for which supplication. To enable and
support a diverse and productive quantum ecosystem, Congress must
ensure that U.S. Government support does not favor one of these
technologies over another because a lack of neutrality could
preemptively stall valuable technological advances.
We are beginning to understand how large and ``good'' a quantum
computer will need to be to break encryption, a possible eventual
threat to cybersecurity. Less well understood is how we will connect
small quantum ``cores'' with hundreds to thousands of qubits to create
systems with hundreds of thousands of qubits. This problem is
challenging because adding more qubits is unlike adding more memory to
a laptop, for example. One doesn't just plug in a hundred more qubits
to get a more powerful system.
In its deliberations on future legislation, such as the renewal of
the 2018 National Quantum Initiative, Congress must call out the need
for domestic development of these ``quantum interconnects'' between
cores. Additionally, this legislation should call out the need for R &
D to understand how several quantum computers might be networked to
create larger systems that could defeat the intent of export controls.
______
Response to Written Question Submitted by Hon. Maria Cantwell to
Henry L. Jones II, Ph.D.
Question. In your written testimony, you emphasize the need to
attract more diverse people into STEM, including through industry
certifications. What workforce difficulties did you see at your company
and what more can universities do to relieve these issues?
Answer. For companies like mine that were competing and growing in
global technology markets, the most significant obstacle to success was
finding, recruiting, signing, and keeping exceptionally talented
technology experts. Capability and availability were the most important
decision factors by far. Diversity was a highly desirable outcome, but
was hampered by our market realities--we would take the qualified hires
we could get.
In addition, many companies have internal programs to improve
recruiting by giving bonuses to its employees who can bring in a
capable hire. This practice tends to result in a workforce with self-
reinforcing demographics, which can significantly impact diversity over
time, especially when compounded with the experience these employees
gain by their initial employment.
Finally, I noticed that often the applicants from communities of
color did not appear to have had opportunities to develop an impressive
work portfolio and useful experience while a student. Rather than
working as an intern applying their newly learned technical skills,
they had been filling non-technical roles as work study students in
university administration offices making copies and shuttling mail
around campus.
The most successful solution that I have is seen for these
difficulties was our creation of a standing internship and work/study
co-op program with nearby universities to connect high potential
juniors and seniors. This allowed the university and my company to
engage students very early in their educational process, to connect
them to role models, to give them insight into the utility of their
classes, and to give them solid work experience to include in their
portfolios to present to future employers. We usually brought these
students in as full-time employees after graduation, but even those
that we did not eventually employ presented positive experiences for
the students and for our company.
Congress could dramatically increase the impact of such programs by
providing cost-share frameworks, creating matchmaking resources, and
supporting any efforts to educate parents in communities of color of
the potential of technical careers for their children and the steps
along the path to get them there.
______
Response to Written Questions Submitted by Hon. Kyrsten Sinema to
Henry L. Jones II, Ph.D.
STEM Education in CHIPS and Science Act: The bipartisan CHIPS and
Science Act authorized $13 billion over five years for the National
Science Foundation (NSF) to put towards Science, Technology,
Engineering, and Mathematics (STEM) education. The legislation directs
the NSF to explore opportunities to engage students from groups
historically underrepresented in STEM fields as well as rural areas to
expand the pipeline to help ensure the United States has the STEM
workforce we need for the Twenty-First Century.
Question 1. How will the CHIPS and Science legislation expand STEM
education and, in turn, America's STEM workforce? Why is it important
for Congress to fund the investments in STEM education made in the
legislation?
Answer. Rural students are generally unlikely to have direct
exposure to cutting-edge technologies and are even less likely to learn
the methods by which they are developed. This exposure comes from
visits to high-tech companies, presentations by role models in their
communities, and projects that give them hands-on experience with the
related tools and concepts. The CHIPS and Science legislation is of
interest to rural states' workforce and industry development because
the expertise areas of individuals, academic teams, and companies that
are needed to create advanced new products and services in rural areas
will be uncommon.
Congress could help address these barriers by supporting activities
that provide educational opportunities to students and to their parents
regarding the career potential of these advanced technologies. Congress
should also ensure that some of the societal and economic forces that
lead to centralization of innovation--personnel and equipment and
infrastructure--are not explicitly or implicitly reinforced by its
policies. The component of the CHIPS Act that emphasizes the
participation of EPSCoR states in the regional technology hubs is an
excellent example of what Congress can do to provide long-term
investment in rural-oriented development of advanced technologies.
Support for workforce development in rural communities focused on
developing and keeping opportunities for advanced technology-related
employment is also something Congress should continue to make a
priority.
Semiconductor Investments: Once the global leader, America now only
accounts for approximately 12 percent of the world's semiconductor
manufacturing. Recognizing this important national security issue, I
worked with my colleagues to ensure the CHIPS and Science Act also
appropriated $52 billion for the Commerce Department to establish
grants to jump-start investments in America's domestic semiconductor
industry. These investments will support tens of thousands of Arizona
jobs at facilities managed by Intel, TSMC, and other Arizona
semiconductor companies.
Question 2. How does the CHIPS and Science Act and the law's
semiconductor investments illustrate the United States' commitment to
being a global leader in this field and what further actions can be
done to expand U.S. leadership in this field?
Answer. The economic drivers that led to the global distribution of
semiconductor expertise and facilities are significantly impacted by
national and international agreements for trade and intellectual
property protection. The CHIPS and Science Act is a substantial action
by the United States to recognize this reality, and to take steps to
set long-term conditions towards a healthy domestic semiconductor
industry.
One area in which Congress might take additional action could be
the sensitive issue of immigration, particularly how we attract,
welcome, and keep the most innovative and energetic individuals,
companies, and communities from around the global semiconductor
industry as American citizens--to integrate them instead of competing
with them. Many of our most successful technology companies (e.g.,
Google) were created by immigrants. Rural areas have very low cost of
living and excellent telecommunications and transport connectivity.
Congress would be wise to implement immigration policies that encourage
global entrepreneurs to move to rural areas.
______
Response to Written Questions Submitted by Hon. Raphael Warnock to
Henry L. Jones II, Ph.D.
Workforce Diversity:
Question 1. What is the importance of fostering greater diversity
within America's science, technology, engineering, and mathematics
(STEM) workforce, and to what extent would failure to diversify our
workforce hinder our national and economic security?
Answer. While the wisdom of diversity is commonplace in investment
circles--the positive economics inherent in diverse portfolios are well
known (even warranting a Nobel Prize for Harry Markowitz)--this same
insight has only recently been recognized by U.S. technology industry
decision makers. Whether the concern is how a lack of diversity might
make the company and its workforce pipeline more fragile by ``putting
all its eggs in one (or just a few) baskets,'' or through the blindness
to potential threats and opportunities due to a limited set of
perspectives, I have witnessed a complete change in mindset over my
career from a disregard for diversity to a highly intentional
appreciation for it at all levels of technology company management.
At this stage, Congress can support this mindset by taking whatever
steps available to it for the further diversification of our STEM
workforce--steps that will be supported by the tech industry. If we as
a nation do not do so, we will undermine our industrial base, our
socioeconomic fabric, and our foundation for future growth.
Question 2. What role do historically black colleges and
universities (HBCUs) and other minority-serving institutions play in
helping to build a diverse workforce? What lessons can universities and
other institutions draw from the success of HBCUs in supporting our
STEM workforce?
Answer. For many years, I was a part of the industry advisory board
created by Jackson State University as it stood up its Computer
Engineering department from scratch in the 2000s, and I watched the
faculty and students rapidly grow professionally and academically. The
obstacles and failures I watched them overcome, as well as the
successes I saw along the way--and the lessons I drew from that
experience--would fill many pages.
The most salient lessons were the critical importance of accessible
role models and high standards, and their impact on student experience
and workforce readiness. Jackson State University, as the HBCU I know
best, had to make many difficult decisions as it created the culture of
its new School of Engineering. HBCUs have an opportunity to promote
world-class excellence in STEM that can have a high-profile impact
across the technology ecosystem. For many engineering tasks, which
impact public safety and well-being in myriad ways, ``good enough'' is
not good enough--standards mean something. It seems to be an
unfortunate truth that even almost 100 years later, the mentality of
the Tuskegee Airmen--that they push themselves to be the best fighter
pilots in the air because of their skin color--is still a societal
dynamic that students of color must deal with. I saw my JSU colleagues
use this challenge as an opportunity to push for a culture of the
highest standards for their students, which raises the bar for STEM
students everywhere no matter their socioeconomic background. I welcome
further discussion that might help Senator Warnock with any efforts on
this vital topic.
Rural Institutions: Georgia is a proud agricultural state, and I am
particularly excited for how advanced technologies will help improve
precision agriculture and otherwise support America's farmers. I am
concerned, however, by the underrepresentation of rural students and
rural colleges and universities among America's leading research
institutions and the effect this will have in fields such as precision
agriculture.
Question 3. What barriers do rural students face in entering the
workforce in advanced technologies such as artificial intelligence and
quantum computing? How can Congress help address these barriers?
Answer. Rural students are generally unlikely to have direct
exposure to cutting-edge technologies in general, and even less likely
to learn the methods by which they are developed. This exposure comes
from visits to high-tech companies, presentations by role models in
their communities, and projects that give them hands-on experience with
technology-related tools and concepts. I agree wholeheartedly with
Senator Warnock that these barriers have the potential to negatively
impact technology topics of interest to rural states, such as precision
agriculture, because the combination of related expertise areas by
individuals, academic teams, and companies that are needed to create
advanced new products and services in rural areas will be uncommon.
Congress could help address these barriers by supporting activities
that provide educational opportunities to students and to their parents
regarding the career potential of these advanced technologies. Congress
should also ensure that some of the societal and economic forces that
lead to centralization of innovation--personnel and equipment and
infrastructure--are not explicitly or implicitly reinforced by its
policies. The component of the CHIPS Act that emphasizes the
participation of EPSCoR states in the regional technology hubs is an
excellent example of what Congress can do to provide long-term
investment in rural-oriented development of advanced technologies.
Question 4. What barriers do rural colleges and universities and
their faculties face in accessing and supporting research related
advanced technologies such as artificial intelligence and quantum
computing? How can Congress help address these barriers?
Answer. The nature of high-growth, high-potential technologies is
that there is often a ``first mover advantage'' in many ways--as
interest grows in a topic, those who have established themselves as the
experts can attract outsized attention and new resources as others join
in and try to get up to speed quickly. That leads to a self-reinforcing
cycle in which the biggest keep getting bigger. If something new
happens to get its start in a rural area, the draw of moving the effort
to the additional resources of an urban area will be significant.
Congress can help with this by recognizing this dynamic (often referred
to as ``power law'' economics) and enacting policies to counter it when
possible. For example, qualified regions or entities could receive
related infrastructure investments (specialized lab facilities,
targeted workforce development programs, etc) that would serve as
anchors to keep innovation tied to specific locations. Support for
workforce development in rural communities focused on developing and
keeping opportunities for advanced technology-related employment is
also something Congress should continue to make a priority.
Global Collaboration and Competition: I have long believed that
America must invest in innovation and advanced technologies to maintain
our global standing. I am deeply alarmed that our global competitors,
such as the Chinese Communist Party, have heavily invested in
technology research and used it to oppress its own people and bolster
autocratic regimes throughout the world. To keep America on the leading
edge, Congress must strike the appropriate level of controls on
innovation and intellectual property to protect our national and
economic security without overly hindering scientific innovation and
international collaboration.
Question 5. What are the key concerns that Congress should keep in
mind as we work to achieve this balance? Please share any examples of
Federal programs that you believe do not strike the right balance.
Answer. Senator Warnock is right to be concerned about this
troubling dynamic within our international competitive environment. In
some cases, various Federal programs and entities might have taken
longer than they should have to recognize this threat, but I do not
currently see any who are acting as if they are unaware of their
responsibility to protect American interests.
The area of greatest concern to Congress should be the sensitive
issue of immigration, particularly how we attract, welcome, and keep
the most innovative and energetic individuals, companies, and
communities from around the globe as American citizens--to integrate
them instead of compete with them. Many of our most successful
technology companies (e.g., Google) were created by immigrants. Rural
areas have very low cost of living and excellent telecommunications
connectivity. Congress would be wise to implement immigration policies
to get global entrepreneurs to move to rural areas, where their impact
will be greater and the social pressures to promote pro-American values
and behaviors will be greater as well.
[all]