NDIA POLICY POINTS ROBOTICS AND AUTONOMOUS SYSTEMS
AI Legislation Languishes in Congress
When the general public hears the term “artificial intelligence,” minds often race to scenes from movies like Terminator or they begin to think about popular rhetoric predicting robotic infiltration of the workplace and the accompanying human job displacement.
These unscientific notions represent the U.S. public’s fascination and unfortunate misunderstanding of these advanced technologies. Complex AI is a transformative advanced technology used to complete tasks traditionally requiring human intelligence. It encompasses an array of computational capabilities often classified as “narrow” or “general” depending on the complexity of the assigned tasks, making it difficult to create one standardized, inclusive definition.
Poor comprehension of AI also extends to Congress. Unlike the public, however, Congress’ policymaking responsibilities make its knowledge deficit harmful to the country, hindering progress on policy, leaving the nation disadvantaged in the global competition for technological supremacy, and possibly creating unanticipated consequences for citizens.
To shrink its knowledge deficit, Congress should establish a joint advisory committee composed of private sector technology experts and top public sector research-and-development professionals by passing the proposed bill, “Fundamentally Understanding the Usability and Realistic Evolution of Artificial Intelligence Act of 2017,” or FUTURE of AI Act. Whether Congress creates legislation or not, AI will continue to develop transforming Americans’ way of life for better or worse.
In December 2017, Rep. John K. Delaney, D-Md., along with eight cosponsors, introduced the bipartisan act into Congress. Along with formally defining AI, the legislation also proposes establishing a federal advisory committee. The committee would consist of two member classes: non-governmental professionals from academia, research, private industry, civil society and labor organizations, and governmental professionals from federal agencies and departments.
The committee would educate and inform policymakers on the implications and functions of AI and recommend policy options. The proposal includes private sector participants because they enjoy substantially more research and funding opportunities in advanced technology.
Collaboration among government personnel creates opportunities to clearly define a standardized government strategy. This partnership also allows the public sector to leverage commercial expertise while avoiding recruiting wars over talent it would almost certainly lose due to salary differentials.
"The government has a responsibility to the nation to incentivize the advancement of AI ..."
Finally, the arrangement benefits the private sector by allowing its experts to collaboratively craft policy impacting their organizations’ strategy, research, technology and product lines.
Since the introduction of the act, congressional interest in AI has surged, producing numerous hearings, white papers and research studies attempting to define a standard policy agenda. As the Center for Strategic and International Studies noted in a recently published report, “an expertise gap persists in technical authorities at the strategic level, with non-specialists often issuing technical decisions.” The report, “Artificial Intelligence and National Security: The Importane of the AI Ecosystem,” concludes that to create strong, effective policies, Congress must become better informed about AI.
Lawmakers’ lack of technological knowledge came to light during the widely watched House and Senate Facebook hearings. Policymakers convened to determine the company’s role in Cambridge Analytica’s controversial use of Facebook users’ personal information. For two days, Congress’ questions to Facebook CEO Mark Zuckerberg portrayed a clear lack of understanding of the platform and its underlying technologies. The result? Zuckerberg dodged the data breach questioning by littering his answers with buzzwords like “machine learning” and “artificial intelligence,” confident those asking the questions would not understand. A video of Sen. Orrin Hatch, R-Utah, asking Zuckerberg how Facebook sustains a business model in which users don’t pay for services went viral after Zuckerberg responded with a chuckle, “Senator, we run ads.”
Lack of appropriate background knowledge prevents members from asking pertinent investigative questions. Without the expertise or understanding, both policy-making and oversight will be inadequate at best and potentially harmful.
In contrast to Congress, the executive branch has already acted. Recognizing the knowledge deficit, President Donald Trump signed an executive order creating a select committee on AI under the National Science and Technology Council. However, this committee only includes government officials, limiting its value. Without expert input from outside the government, regulations and policy will lag technological innovation, either slowing growth and hindering the economy or failing to ensure the U.S. implements effective guidelines for new capabilities. While this action demonstrates progress, it is insufficient.
Lawmakers introduced the FUTURE of AI Act of 2017 more than a year ago; while it languishes in Congress, the United States and its adversaries continue to push technological boundaries with little input or oversight from the U.S. government. Congress needs to prioritize passage of the act to effectively legislate and regulate AI in a manner that fosters innovation, grows the economy, protects U.S. citizens from unanticipated consequences of ungoverned research and development, all the while keeping the United States as the leader in this transformative technology.
The government has a responsibility to the nation to incentivize the advancement of AI in the right direction while ensuring the United States is prepared for the technological demands of tomorrow.
Alexandra Berge is a junior fellow at NDIA’s policy division.
Topics: Robotics, Robotics and Autonomous Systems, Infotech, Cyber, Cybersecurity