Over the last two months, the United States legislature has introduced three new bills intended to establish a Federal Advisory Committee on the rapidly-evolving field of artificial intelligence (AI) and to analyze and report on the impact and growth of the technology.
FUTURE of AI Act
On December 12, 2017, Rep. John Delaney (D-MD) and Sen. Maria Cantwell (D-WA) introduced identical bills in the House (as H.R. 4625) and Senate (as S. 2217). The twin bills were entitled the "Fundamentally Understanding The Usability and Realistic Evolution of Artificial Intelligence Act of 2017" -- or FUTURE of AI Act.
One aim of the bills is to better understand how AI might maximally benefit the economic prosperity and social stability of the United States. The bills would also direct the Secretary of Commerce to establish a Federal Advisory Committee, which would be tasked with providing guidance on the development and implementation of artificial intelligence. Among other roles, the Advisory Committee is to promote a "climate of investment and innovation," "optimize the development of [AI]," support the "unbiased development and application of [AI]," and "protect the privacy rights of individuals."
The Advisory Committee is asked to provide advice to the Secretary of Commerce with regard to several specific topics including:
(A) The competitiveness of the United States, including matters relating to the promotion of public and private sector investment and innovation into the development of artificial intelligence.
(B) Workforce, including matters relating to the potential for using artificial intelligence for rapid retraining of workers, due to the possible effect of technological displacement.
(C) Education, including matters relating to science, technology, engineering, and mathematics education to prepare the United States workforce as the needs of employers change.
(D) Ethics training and development for technologists working on artificial intelligence.
(E) Matters relating to open sharing of data and the open sharing of research on artificial intelligence.
(F) International cooperation and competitiveness, including matters relating to the competitive international landscape for artificial intelligence-related industries.
(G) Accountability and legal rights, including matters relating to the responsibility for any violations of laws by an artificial intelligence system and the compatibility of international regulations.
(H) Matters relating to machine learning bias through core cultural and societal norms.
(I) Matters relating to how artificial intelligence can serve or enhance opportunities in rural communities.
(J) Government efficiency, including matters relating to how to promote cost saving and streamline operations.
The Advisory Committee is additionally required to conduct a study on various intersections between AI and society, including analyzing the effects of AI on the economy, workforce, and competiveness of the United States. The study also seeks to identify and eliminate bias in AI algorithms, identify potential "harmful outcomes," and contemplate the incorporation of ethical standards in AI development. Furthermore, within 540 days of enactment of the Act, the Advisory Committee is required to provide a report based on the study to the Secretary of Commerce and Congress.
The bills direct that the Advisory Committee shall include 19 voting members appointed by the Secretary of Commerce. Specifically, the voting members will be selected from "diverse" geographical locations within the United States and include: five members from academia or the AI research community, six from private industry, six from civil society, and two from labor organizations.
One of the notable features of the twin House and Senate bills is the broad range of technologies being considered under the term "artificial intelligence." The bills both define AI as:
(A) Any artificial systems that perform tasks under varying and unpredictable circumstances, without significant human oversight, or that can learn from their experience and improve their performance. Such systems may be developed in computer software, physical hardware, or other contexts not yet contemplated. They may solve tasks requiring humanlike perception, cognition, planning, learning, communication, or physical action. In general, the more human-like the system within the context of its tasks, the more it can be said to use artificial intelligence.
(B) Systems that think like humans, such as cognitive architectures and neural networks.
(C) Systems that act like humans, such as systems that can pass the Turing test or other comparable test via natural language processing, knowledge representation, automated reasoning, and learning.
(D) A set of techniques, including machine learning, that seeks to approximate some cognitive task.
(E) Systems that act rationally, such as intelligent software agents and embodied robots that achieve goals via perception, planning, reasoning, learning, communicating, decision making, and acting.
H.R. 4625 and S. 2217 also contemplate "artificial general intelligence," which they define as "a notional future artificial intelligence system that exhibits apparently intelligent behavior at least as advanced as a person across the range of cognitive, emotional, and social behaviors." Yet further, the bills consider "narrow artificial intelligence," or artificial intelligence systems that address specific applications such as playing strategic games, language translation, self-driving vehicles, and image recognition.
AI Jobs Act of 2018
On January 18, 2018, Rep. Darren Soto (D-FL) introduced H.R. 4829 entitled the "AI Jobs Act of 2018". The bill promotes a "21st century artificial intelligence workforce," and focuses on training and retraining American workers in light of the possible effects of AI on the workforce and human worker demand.
H.R. 4829 requires the Secretary of Labor to undertake an AI study to determine its potential impacts on the workforce. A report based on the study is to include:
(1) An outline of the specific data, and the availability of such data, necessary to properly analyze the impact and growth of artificial intelligence.
(2) Identification of industries that are projected to have the most growth in artificial intelligence use, and whether the technology will result in the enhancement of workers' capabilities or their replacement.
(3) Analysis of the expertise and education (including computer science literacy) needed to develop, operate, or work alongside artificial intelligence over the next two decades, as compared to the levels of such expertise and education among the current workforce.
(4) Analysis of which demographics (including ethnic, gender, economic, age, and regional) may experience expanded career opportunities, and which such demographics may be vulnerable to career displacement, due to artificial intelligence.
(5) Any recommendations to alleviate workforce displacement, prepare future workforce members for the artificial-intelligence economy, and any other relevant observations or recommendations within the field of artificial intelligence.
In preparing the report, the Secretary of Labor is tasked with conducting a series of public hearing or roundtables with various organizations, such as industrial stakeholders, heads of Federal agencies, local educational agencies, institutions of higher education, workforce training organizations, and National Laboratories.
While some language is common, the new House bill defines "artificial intelligence" more succinctly than the earlier twin bills. Specifically, H.R. 4829 characterizes AI as systems that:
(A) think like humans (including cognitive architectures and neural networks);
(B) act like humans (such as passing the Turing test using natural language processing, knowledge representation, automated reasoning, and learning);
(C) think rationally (such as logic solvers, inference, and optimization);
(D) act rationally (such as intelligent software agents and embodied robots that achieve goals via perception, planning, reasoning, learning, communicating, decision-making, and acting); or
(E) automate or replicate intelligent behavior.
If enacted, the Acts and their resulting reports will undoubtedly lead to further discussion regarding artificial intelligence and provide multiple perspectives on how AI might impact society for better and, perhaps, for worse. Additionally, as the bills target the Departments of Commerce and Labor separately, it will also be interesting to see how the different organizations read and react based on their respective AI studies.
For additional information regarding this topic, please see:
• H.R.4625 - FUTURE of Artificial Intelligence Act of 2017
• S.2217 - FUTURE of Artificial Intelligence Act of 2017
• H.R.4829 - AI JOBS Act of 2018
• "Lawmakers introduce bipartisan AI legislation," The Hill, December 12, 2017
• "Microsoft Sees Need for AI Laws, Regulations," Industry Week, January 18, 2018
• "Forget Killer Robots—Bias Is the Real AI Danger," MIT Technology Review, October 3, 2017