Senator Schumer’s Proposal for AI Legislation Is on the Right Track

Senate Majority Leader Schumer (D-NY) launched a two-pronged proposal this week for a U.S. policy approach to AI. The first part lays out five principles to guide members of Congress on what AI policies should achieve, while the second part proposes a new process to better translate legislative ideas into action. While the proposal lacks a clear call to promote widespread AI adoption, the Senator offers a sensible path to generating AI policy that remains innovation-friendly and balanced.

The Senator’s proposed framework for AI legislation, called the SAFE Innovation Framework, consists of five key policy objectives: 1) security, meaning AI policies should safeguard U.S. national security from foreign adversaries that may use the technology in malicious ways and secure U.S. economic well being from AI-related job loss; 2) accountability, meaning AI policies should ensure AI systems are deployed responsibly by addressing concerns about bias and misinformation, protecting intellectual property, and addressing liability; 3) foundations, meaning AI policies should ensure AI systems are developed and deployed in ways that promote democratic values, including elections; 4) explain, meaning AI policies should determine what information the public needs to know about an AI system, and when; and 5) innovation, meaning AI policies should support U.S.-led innovation in AI technologies and maintaining U.S. leadership in the technology.

Most of these principles are reasonable, but they aren’t perfect. For example, in his remarks announcing this proposal, Senator Schumer argued for “measures to prevent job loss.” Clearly, policymakers should provide help so that workers impacted by automation-related job losses have a strong social safety net and can pursue training for new jobs. But AI policy should be focused on how to effectively harness the productivity gains AI offers by boosting labor productivity, ensuring that workers leverage those gains in higher wages, and providing adjustment support for those left behind, not preventing any job losses. Indeed, the Senator should recognize that creating the right policies to support AI-enabled work is an opportunity to help address middle class stagnation, rather than assuming the technology will cause it.

The second part of his policy proposal is to “create a new legislative approach for translating this framework into legislative action.” Normally Congress uses hearings to gather information, examine issues, and make decisions on legislation. But as Senator Schumer explained, hearings on their own are not the most effective process for crafting AI legislation because they tend to include only a few witnesses and allow for only a few questions from members of Congress, which can be insufficient for AI issues that are technically complex, constantly evolving, and benefit from broad expert input. Instead, the Senator wisely proposes to create “AI insight forums,” which would be a series of panels focused on specific AI issues and would include “top AI developers, executives, scientists, advocates, community leaders, workers, [and] national security experts.” This initiative will be co-led by Senators Schumer, Heinrich (D-NM), Young (R-IN), and Rounds (R-ND) to ensure it remains bipartisan. Senator Schumer wants members of Congress to listen to these experts and translate their insights into legislative action.

However, there was one word missing from Senator Schumer’s proposal: adoption. While the Senator is right that AI does indeed offer enormous societal and economic benefits in sectors such as health care, transportation, and education, it is important to recognize these benefits will not be realized by only improving AI development. The United States also needs a multipronged national AI adoption strategy to ensure these opportunities are translated into all the areas where they can make a positive difference in people’s lives. It is therefore critical that Congress focus on crafting policies that accelerate the public-sector adoption of AI by addressing challenges related to acquisition, funding, and oversight as well as industry adoption of AI by supporting sector-specific AI strategies.

The Senator should also heed the warning from his Democratic colleague Representative Ted Lieu (D-CA) on the pitfalls of Congress moving too quickly in this area. While it is indeed critical that the United States is an AI leader and that policymakers and the public feel secure that the risks of AI are addressed, if Congress is hasty in passing AI laws, it could cement flawed legislation that may be more difficult to fix than the issues at hand, as the EU will likely learn to their dismay. Indeed, one of the biggest risks with Schumer’s fast-track proposal is that in an environment where many argue, without evidence, that AI presents an existential risk to humanity, it will be nearly impossible to create reasonable policy solutions. Worse, it can legitimize ill-conceived policy solutions simply because they replicate the status quo consensus in moments of hype, even if this consensus is wrong and overwrought. Ideally, rather than rush to create new laws, Congress should start by ensuring existing laws and regulations, such as nondiscrimination laws that already apply to AI systems, are rigorously enforced and conduct a gap analysis to identify any shortcomings. Where necessary, it should carefully craft and evaluate regulatory proposals for AI to ensure they are sufficiently targeted and do not harm innovation.

Overall, Senator Schumer has offered a reasonable path forward for Congress to pursue as it seeks to better understand AI. He rightly recognizes the significant benefits AI can provide and is taking an innovation-friendly approach that will set the nation up for success. Unfortunately, many anti-AI voices will likely flood this process with calls for stringent, European-style restrictions on AI across broad swaths of the U.S. economy. If the United States wants to remain ahead of China on AI, it will need to resist such calls.