The Good, The Bad, and the Unanswered in Sen. Schumer’s AI Roadmap

On May 15, Senate Majority Leader Chuck Schumer (D-N.Y.) and the Bipartisan Senate Artificial Intelligence (AI) Working Group released a report, entitled Driving U.S. Innovation in Artificial Intelligence: A Roadmap for Artificial Intelligence Policy in the U.S. Senate. As the title indicates, this report is intended to serve as a roadmap for AI policy in the United States in the short and long term. It highlights policy priorities, questions to answer, and benefits and concerns on the development of AI.

The document provides various insights on how the Senate, and perhaps Congress at large, will tackle the rise of AI. It seeks to provide more clarity on what questions cause the most consternation in Congress. It also details what AI-related legislation might come in the future.

Thankfully, the document departs from the predominantly negative and apocalyptic narrative that has dominated discussions of AI policy recently. However, while the report does bring clarity on various issues, it also leaves many questions still unanswered. Overall, the report is a mixed bag of good ideas, bad ideas, and unanswered questions:

The Good

The Report Takes a Step Away from the Predominant “Existential Risk” Narrative

As it has been noted, the working group largely avoided apocalyptic “existential risk” arguments surrounding AI technology and takes a much more optimistic approach. The report acknowledges that this technology will have a net positive effect on humanity. It qualifies that certain points of concern must first be tackled. Instead of focusing on restrictions, the report largely focuses on how to ensure AI develops as intended. Focusing on how to get AI deployment right, instead of just preventing it from going wrong, is a valuable first step. This will ensure AI regulations have a pro-growth approach that maximizes the technology’s potential for good.

Initiatives That Invest Money to Save Money

The roadmap suggests that the federal government should significantly increase AI-related expenditures in the short-term. Due to the broad range of the laundry list of items to spend on, most of these expenditures would not be a good use of taxpayers’ money. Instead of direct spending on research and development (R&D), Congress should reinstate full expensing of private R&D investment. However, some of these suggested expenditures could be valuable investments that will make sure that the adoption process of AI in government is done in an efficient, safe, and reliable manner, and ultimately protect taxpayers in the future. For example, the roadmap suggests government R&D investments in updating infrastructure and ensuring that government has adequate data analytics software. It also stipulates investments to integrate AI technology in government services, ensure government databases are “AI-ready,” and to leverage AI to find inefficiencies in the U.S. code, federal rules, and procurement programs.

Commitment to Adapt Existing Laws Instead of Passing New Legislation

 Concerns over how AI could impact various parts of the economy have spurred waves of legislation. Lawmakers aim to regulate the use of AI in “high impact” areas. However, as others have pointed out, most of the behaviors that policymakers are concerned about are mostly covered by existing laws. Before enacting new, potentially stifling regulations, it is important that policymakers ensure that existing laws are correctly applied. If necessary, adjustments can be made before overhauling the whole landscape. AI-specific legislation should be used as a last resort, only when existing regulations or precedent have proven insufficient or inapplicable. Fortunately, the report has adopted this similar position, explicitly expressing its belief that “existing laws, including related to consumer protection and civil rights, need to consistently and effectively apply to AI systems and their developers, deployers, and users.” Additionally, it claims that additional regulations should come in a case-by-case basis only when any gaps have been properly identified.

The Bad

AI Industrial Policy

Unfortunately, the report reinforced the idea that industrial policy will continue to be a tool that policymakers are eager to use. As mentioned above, the report creates a broad list of expenditures that it believes the federal government should be making in the AI field. For example, the report reaffirms its commitment to the CHIPS Act, pushing for additional funding in some areas that are slated to but have yet to receive funding. Despite the repeated failures of industrial policy, the policy consensus around federal spending in AI has not waned.

The Unanswered

Will AI Bear the Liability for User-Generated Content?

While the report has resolved various doubts and set a concrete roadmap for the future of AI policy, there were a set of questions that were largely set unanswered. A key unanswered question is where legal liability will rest for user-generated content using AI tools. The report paid some lip service to this question but does not give a definitive answer on this topic. As some proposals to strip AI of Section 230 protections have already surfaced in the Senate, there is a cause for worry that AI will not benefit from a basic protection that is largely responsible for the success of the U.S. tech industry. 

The Future of Work with AI

The working group recognized that AI has the potential to bring tremendous economic growth through an AI-enabled productivity boost. However, it also voiced concerns over how AI and automation could lead to a potential displacement of workers. The report claims that the working group should take measures to ensure that “American workers are not left behind,” but its list of recommendations is largely ambiguous. This ambiguity leaves a dangerous door open for overreach and onerous regulations that could limit the American economy from implementing AI tools in order to protect any potential job losses – or to create entirely new jobs.

Government Adoption of AI

Despite the report’s overall favorable tone towards AI, it largely left the questions regarding government adoption of AI unaddressed. While it mentioned that there should be R&D investments to ensure that government has the necessary infrastructure in place for adoption of AI, it largely lacked any mention of adoption strategies, training policies, or the need for any guardrails for government use.

Overall, the report shows that the Senate has been able to drown out some of the noise from the AI doomsayers. It primarily focuses on how to maximize the potential of AI to do good. While it provided various key answers that required more clarity, it left crucial questions unanswered. Hopefully this positive sentiment translates into future policies that leverage the technology for a pro-growth, pro-innovation agenda.