AI strategy, objectives, and metrics plans

(a)

Amendment

Chapter 3 of title 5, United States Code, is amended by adding at the end the following new subchapter:

III

Artificial Intelligence

321.

Definitions

In this subchapter:(1)

Agency

The term agency has the meaning given that term in paragraph (1) section 3502 of title 44, United States Code, except that term shall not include independent regulatory agencies (as defined in paragraph (5) of such section).(2)

Appropriate congressional committees

The term appropriate congressional committees means the Committee on Oversight and Reform of the House of Representatives and the Committee on Homeland Security and Governmental Affairs of the Senate.(3)

Artificial intelligence

The term artificial intelligence has the meaning given that term in section 238(g) of the John S. McCain National Defense Authorization Act for Fiscal Year 2019 (Public Law 115–232; 10 U.S.C. 2358 note).(4)

Director

The term Director means the Director of the Office of Management and Budget.(5)

Reliability

The term reliability means, with respect to the artificial intelligence, the extent to which—(A)data collected for use by the artificial intelligence is a valid measure of the underlying concepts addressed in the objectives of using the artificial intelligence;(B)the safety, security, and robustness has been tested and assured across each phase of development and use of the artificial intelligence according to defined objectives;(C)data records and fields are present and sufficiently populated; and(D)data records reflect the actual underlying information.(6)

Representativeness

The term representativeness means the extent to which data sets used for training the artificial intelligence is representative of the population or operational environment to be assessed by the artificial intelligence.

322.

AI Strategy, Objectives, and Metrics Plan

(a)

AI Strategy, Objectives, and Metrics Plan

The head of each agency shall establish an AI Strategy, Objectives, and Metrics Plan that contains strategies, objectives, and metrics for the trustworthy adoption of artificial intelligence by the agency to better achieve the mission of the agency to serve the people of the United States, including the following:(1)Defined roles, responsibilities, and delegation of authority for the use and oversight of artificial intelligence used by the agency, including oversight of compliance with relevant laws, regulations, standards, and guidance.(2)Defined values, ethics, and principles to foster public trust and responsible use of artificial intelligence by the agency.(3)Procedures—(A)to ensure the security and privacy of data used by artificial intelligence used by the agency; and (B)for human supervision of, and accountability for, artificial intelligence used by the agency. (4)For each artificial intelligence used by the agency at the time the plan is established—(A)performance objectives and accompanying metrics that clearly reflect or convey the intended purpose of the artificial intelligence; (B)a process—(i)for the continuous or routine monitoring and assessment the performance of the artificial intelligence against the objectives and metrics established for the artificial intelligence under subparagraph (A) to ensure that the artificial intelligence functions as intended and carries out the goals and objectives of using the artificial intelligence; and(ii)that includes a plan to systematically identify, analyze, and mitigate any risks associated with the artificial intelligence; and(C)documentation demonstrating that the artificial intelligence relies on high quality, reliable, representative data, including—(i)the sources and origins of data used to train each artificial intelligence; and(ii)assessments of the reliability, quality, and representativeness of—(I)data used to train the artificial intelligence;(II)attributes used to categorize data; (III)data variables used by the artificial intelligence; and(IV)interconnectivities and dependencies of data streams that operationalize the artificial intelligence; and (iii)a description of the range of data and model drift that is acceptable for each artificial intelligence.(5)Performance objectives and accompanying metrics, for each opportunity identified by the agency to incorporate the use of artificial intelligence into the platforms, processes, and operations of the agency. (6)A description of how artificial intelligence used by the agency are monitored, including a description of any actions taken to correct the functioning or performance of the artificial intelligence.(7)A description of how information on the design, operation, limitations, and updates of the artificial intelligence is made available to the public. (b)

Relationship to Executive Order 13859, memoranda

A plan developed under subsection (a) by an agency shall be consistent with—(1)section 3 of Executive Order 13960 (89 Fed. Reg. 13859) titled Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government, issued on December 8, 2020; and(2)the memorandum issued by the Director to the agency under section 104 of the AI in Government Act of 2020 (Public Law 116–260; 40 U.S.C. 11301 note). (c)

Reports

(1)

Initial report

Not later than 180 days after the memorandum required by section 104(a) of the AI in Government Act of 2020 (Public Law 116–260; 40 U.S.C. 11301 note) is issued, the head of each agency shall submit to the Director, the Inspector General of the agency, and the appropriate congressional committees a report that contains—(A)the AI Strategy, Objectives, and Metrics Plan established by the agency under subsection (a); (B)an inventory (including a catalogue of model and non-model components, specifications, and parameters) of all artificial intelligence in use by the agency, in accordance with the guidance issued under paragraph (3); and (C)a description and assessment of opportunities to incorporate the use artificial intelligence into the platforms, processes, and operations of the agency to enhance the effectiveness of the agency in accomplishing the mission, operations, and programs of the agency.(2)

Progress reports

In 2023, and every two years thereafter until 2033, the head of each agency shall submit to the Director and the appropriate congressional committees a report—(A)on the implementation of the AI Strategy, Objectives, and Metrics Plan established by the agency under subsection (a); and(B)that identifies and assesses any opportunities for, and risks any associated with, the use, or potential use, of artificial intelligence by the agency.(3)

Agency inventory guidance

Not later than 30 days after the date of the enactment of this Act, the Chief Information Officers Council shall issue, and make publicly available, guidance that includes the criteria, format, and mechanisms for the inventory required to be included in the report under paragraph (1). (4)

Reporting reciprocity

The reports required by paragraphs (1) and (2) shall fulfill the requirements of the AI in Government Act of 2020 (Public Law 116–260; 40 U.S.C. 11301 note) and subsections (b) and (c) of section 5 of Executive Order 13960 (89 Fed. Reg. 13859) titled Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government, issued on December 8, 2020, and shall include all information required by such provisions.(d)

Consultation

In developing a plan required to be submitted under subsection (a) and the reports required by subsection (c), the head of the agency shall consult with the National Cyber Director and the heads of the National Artificial Intelligence Initiative Office, the National Artificial Intelligence Advisory Committee, the Artificial Intelligence Center of Excellence. (e)

Classified annex

The reports required by paragraphs (1) and (2) of subsection (b) shall be made publicly available upon submission to the entities described in such paragraphs, but such reports may include a classified annex.(f)

Rule of construction

Nothing in this subchapter may be construed as authorizing the use by agencies of, or an appropriation for, artificial intelligence.

(b)

Clerical amendment

The table of contents for chapter 3 of title 5, United States Code, is amended by adding after the item relating to section 315 the following:

Subchapter III—Artificial Intelligence

321. Definitions

322. AI strategy, objectives, and metrics plans required