In the final months of every Congress, lawmakers must weather a deluge of legislative proposals, as those about to lose their majority or leave office rush to ram through their pet projects. Reports indicate that Sen. Chuck Schumer (D-N.Y.), the Senate majority leader until January, is readying a package on artificial intelligence (AI). AI has proven a difficult topic for the 118th Congress, which has fielded myriad bills from various members, each containing widely diverging approaches.
While its details remain unfinalized (publicly, at least), the Schumer package seems likely to entrench the Progressive theory of government — i.e., governance by bureaucratic dictate, not democratic political negotiation — in the regulation of the digital world. To regulate AI, Schumer will likely propose investing administrative bodies with significant powers, largely without congressional oversight. Two specific proposals seem particularly worrisome.
First, the CREATE AI Act would codify the National Artificial Intelligence Research Resource (NAIRR), “a cloud computing resource that will democratize development and use of [AI]” (as put by a Senate brief). While the NAIRR could benefit small developers, who often lack the extensive resources enjoyed by large AI firms, it brings its share of risks. Control of the NAIRR’s standards would likely fall to special interests or ideologically driven political factions, which would distort its operations and, by extension, the trajectory of AI development.
The NAIRR’s short history gives little reason for confidence. A January 2023 report from the Biden administration, which serves as the CREATE AI Act’s source material, calls for “advancing diversity, equity, inclusion, and accessibility (DEIA) in all aspects of the NAIRR, including operations.” This could, for example, compel developers seeking NAIRR grants to tailor their projects to gratify the biases of ideologically blinkered regulators.
Multiple bills, including the AI Advancement and Reliability Act, attempt to give a statutory imprimatur to the AI Safety Institute, a body created pursuant to an executive order from President Joe Biden. The Institute could prove another vector bureaucratic lawmaking, safe from congressional oversight — a particularly acute danger if the Institute is to be staffed by individuals with dubious judgement, as it is now. “Initially created as a standards-setting body…the institute could be poised to become a de-facto regulatory authority,” argues Kristian Stout, the director of innovation policy at the International Center of Law and Economics. Experience has shown the propensity of jawboning and soft law to harden, bending industry to regulators’ will without the trouble of navigating traditional legislative or rulemaking channels. “This could create an avenue for AI safety alarmists to shape restrictive policies through ‘voluntary’ standards and guidelines that effectively become mandatory through market pressure, government contracting, and judicial deference,” Stout continues. The power of “suggestion,” when the state does the suggesting, is the looming threat of coercion.
Stout also worries that federally endorsed datasets might exert a gravitational force on developers, prompting industry to abandon other options. “Instead of fostering a diverse ecosystem of AI research and development, it could result in homogenization, where every AI model is trained on the same narrow datasets, limiting breakthroughs that might arise from alternative approaches,” he argues. It is impossible to avoid the fact that a regulator’s endorsement of one practice ineluctably discriminates against competing practices.
Modern Americans may not live in Woodrow Wilson’s world, but they are largely governed by the Constitution as he envisioned it. Progressive politicians, including Wilson and President Franklin D. Roosevelt shifted vast powers from Congress to administrative agencies. Today, in many policy arenas, such agencies as the Federal Trade Commission (est. 1914), the Securities and Exchange Commission (est. 1934), and the Environmental Protection Agency have displaced Congress as the primary policymakers in their respective arenas, designing laws under which Americans, despite having never voted for those laws’ authors, must live and work.
To those who prefer the Founders’ Constitution, which intends the people’s elected representatives to drive policymaking, the prospect of bureaucratizing the governance of the digital world is intolerable. There is, of course, a place for the executive branch in crafting tech policy. The issues in question often require technical expertise beyond Congress’s capacity. But neither should Congress erect regulatory apparatus with which bureaucrats might dictate sweeping rules for innovators. At either the NAIRR or the AI Safety Institute, small cadres of interested lobbyists or progressive ideologues could warp policy to advance their own aims, shielded from congressional direction and oversight.
“A dependence on the people is, no doubt, the primary control on the government,” James Madison wrote in Federalist No. 51. The erasure of much of that dependence — accomplished by transferring Congress’s constitutional prerogatives to the administrative state — has stained the last century of American history. The Progressive approach ought not be applied to the digital world.
David B. McGarry is a policy analyst at the Taxpayers Protection Alliance.