[Congressional Record Volume 169, Number 195 (Tuesday, November 28, 2023)]
[Senate]
[Pages S5620-S5621]
From the Congressional Record Online through the Government Publishing Office [www.gpo.gov]
Artificial Intelligence
Mr. THUNE. Mr. President, artificial intelligence--or what is
referred to as AI--has been with us for quite some time now. And
whether it is the chatbot providing help on Amazon, or personalized
recommendations on Netflix, or the algorithms that are curating your
social media feeds, these days, most of us interact with artificial
intelligence on a daily basis. But as the release of ChatGPT to the
public last year demonstrated, artificial intelligence is about to take
a giant step forward.
The AI applications I have mentioned, like chatbots and personalized
recommendations, are examples of so-called narrow AI, which is AI
trained to perform specific tasks. But ChatGPT is an example of the
next generation of AI--artificial intelligence systems set up to
imitate the human brain and produce original content based on the
assimilation of vast sets of data.
This next generation of AI--so-called foundation models which
underpin systems like ChatGPT--offers tremendous possibilities:
advances in medicine, in farming, in manufacturing, the automation of
routine tasks, new ways to manage infrastructure, better and more
resilient supply chains, advances in national defense. And the list,
literally, goes on and on.
But as with any sophisticated technology, this next generation of AI
presents risks as well. And those risks are heightened by the enormous
capabilities of AI and the potential for this technology to pervade
every corner of our society. And our goal needs to be encouraging the
promise of AI while putting safeguards in place to minimize potential
dangers.
The light-touch approach the United States has taken to internet
regulation is a good model to follow as we approach AI regulation. The
explosive growth of internet innovation in the United States is, in
large part, a result of the fact that the government has not weighed
down this sector of the economy with heavyhanded regulation. And we
should maintain a similarly light touch when it comes to AI to
encourage innovation and keep the United States at the forefront of the
next generation of artificial intelligence.
Leadership in AI will benefit our economy. And there are also serious
security reasons why staying at the forefront of the AI revolution is
important. There is no question that AI will come to play an important
role in national defense, and falling behind adversaries--like the
Chinese Communist Party--could put our country at a serious
disadvantage when it comes to our national security. So we need to
start establishing some basic rules of the road that will allow AI
innovation to flourish, while at the same time minimizing the dangers
that it presents.
The race to regulate AI has already started. President Biden has
released a sweeping Executive order that empowers multiple government
Agencies and Departments to regulate all AI systems, and even the
algorithms that recommend our next movie on Netflix or remind us that
we need to order more paper towels.
And internationally, the European Union has continued to press
forward with a heavyhanded regulatory regime. It is time for Congress
to ensure that innovation in the United States continues. Regulating AI
by Executive order is not the way to go about things.
Even if the President's Executive order on AI weren't overly broad
and heavyhanded, Executive orders are, by their very nature, not
permanent, since they have the potential to be reversed or amended at
any time--and stand a good chance of being reversed or amended when a
new administration comes into office. This creates uncertainty for
companies, which can stunt expansion and innovation.
The right way to approach AI regulation is to pursue a bipartisan,
nationwide approach in Congress that will protect innovation while
putting in place the necessary safeguards for the riskiest applications
of this technology.
To that end, shortly before Thanksgiving, I introduced bipartisan AI
legislation with Senator Klobuchar and several of our Commerce
Committee colleagues from both parties. Our bill is intended to
establish some basic rules of the road for artificial intelligence
while protecting the ability of companies to innovate and advance this
technology.
Our bill focuses on two things: transparency for consumers and a
tiered, risk-based framework for oversight of the highest impact
applications of AI. On the transparency front, our bill would require
any large-scale internet platform that uses generative AI to create
content to clearly inform consumers of that fact. One of the risks
presented by generative AI is the difficulty distinguishing AI-produced
content from human-produced content.
Now, that may not be a huge issue if the content we are talking about
is an amusing meme, but it is a real issue if a consumer is trying to
figure out whether information or an image is real or whether it has
been generated by AI. So requiring transparency about whether content
is being produced--or at least partially produced--by generative AI
needs to be a priority.
The second part of our bill deals with high-impact and critical-
impact AI, that is, AI applications used to make significant
applications in particularly high-risk sectors. Our bill establishes a
two-tiered system for overseeing these applications.
[[Page S5621]]
Critical-impact AI applications--for example, like those used to make
significant decisions in the operation of critical infrastructure--
would be required to self-certified compliance with testing evaluation,
validation, and verification standards.
What does all that mean? High-impact AI applications will be subject
solely to transparency reporting requirements. Importantly, this part
of the bill is carefully tailored to apply only to AI applications
making complex decisions in high-risk sectors and is meant to respond
directly and narrowly to the recent leap in capabilities of foundation
models that power them.
I believe that the bill Senator Klobuchar and I have introduced is
the right first step when it comes to AI technology. Unlike the White
House's Executive order, our bill doesn't instantly assume that
artificial technology is bad and that it should be subject to
heavyhanded government intervention, nor does our legislation rush us
into regulations before we have a clear idea of what aspects of this
technology need to be regulated and in what way.
Instead, our bill puts in place guardrails to mitigate the dangers
posed by the highest-impact AI application, while leaving American
innovators and entrepreneurs free to move forward with innovation.
I am grateful to Senator Klobuchar and our other cosponsors for
working with me on this bill, and we will continue to welcome ideas to
further improve our legislation. Legislation on this issue of this
magnitude calls for the deliberation of the committee process and
regular order consideration, and I will work to ensure that we take it
up in the Commerce Committee in the coming months.
This bill will probably not be the last bill that Congress needs to
consider when it comes to AI, but I believe it is the right place for
us to begin. And I look forward to working with colleagues from both
parties to getting this bill through Congress and across the finish
line.
I yield the floor.
I suggest the absence of a quorum.
The ACTING PRESIDENT pro tempore. The clerk will call the roll.
The senior assistant legislative clerk proceeded to call the roll.
Mr. MORAN. Mr. President, I ask unanimous consent that the order for
the quorum call be rescinded.
The PRESIDING OFFICER (Mr. Padilla). Without objection, it is so
ordered.