Senate study proposes ‘at least’ $32 billion per year for AI programs

A long-running Senate task force has released its policy recommendation for federal AI funding: $32 billion a year, covering everything from infrastructure to grand challenges to national security risk assessments.

This “road map” is not a bill or a detailed policy proposal, but it nevertheless gives a sense of the scale that legislators and “stakeholders” face whenever they consider reality – although the likelihood of this happening in an election year is low. infinitely small.

In a final report released by the office of Senate Majority Leader Chuck Schumer (D-N.Y.), the bipartisan task force identifies the most important areas of investment to keep the United States competitive with its foreign rivals.

Here are some key elements of the roadmap:

• “An intergovernmental R&D effort on AI, including relevant infrastructure,” which means getting DOE, NSF, NIST, NASA, Commerce and half a dozen other agencies and departments to format and share data in an AI-friendly way. In some ways, this seemingly simple task is the most difficult of all and will likely take years to accomplish.
• Fund U.S. work on AI hardware and software at the semiconductor and architectural level, both through the CHIPS Act and elsewhere.
• Further fund and develop the national AI research resource, still in its infancy.
• “The Grand Challenges of AI” to drive innovation through competition in “applications of AI that would fundamentally transform the process of science, engineering or medicine, as well as in the fundamental subjects of the design of secure and efficient software and hardware”.
• “Support AI preparedness and cybersecurity” in elections, particularly to “mitigate AI-generated content that is objectively false, while protecting First Amendment rights.” Probably harder than it looks!
• “Modernize the federal government and improve government service delivery” by “updating IT infrastructure to use modern data science and AI technologies and deploying new technologies to detect inefficiencies in U.S. code, federal rules and procurement programs. I understand what they’re saying here, but that’s a lot to do for an AI program.
• Lots of vague but important defense-related things like “assessment and mitigation of chemical, biological, radiological, and nuclear (CBRN) AI-enhanced threats by DOD, Department of Homeland Security (DHS), DOE and other relevant agencies.”
• Examine the “regulatory gap” in finance and housing, where AI-enabled processes can be used to further marginalize vulnerable groups.
• “Consider whether other potential uses of AI should be extremely limited or prohibited. » After a section on potentially dangerous things like AI-based social scores.
• Legislation prohibiting AI-generated child sexual abuse material and other non-consensual images and media.
• Ensure NIH, HHS, and FDA have the tools necessary to evaluate AI tools in medical and healthcare applications.
• “Establish a consistent approach to public transparency requirements for AI systems,” private and public.
• Improve the general availability of “content provenance information”, i.e. training data. What was used to make a model? Is your use of the model used to train it further? And so on. AI creators will fight this tooth and nail until they can sufficiently clean up the stores of ill-gotten data they used to create today’s AIs.
• Consider the risks and benefits of using private AI versus open source AI (if the latter ever exists in a scalable form).

You can read the full report here; there are many more bullet points where the above comes from (a longer list than I had planned to write). No budget figure is suggested.

Given that the next six months will be spent mostly on election-related prevarication, this document serves more to anchor many general ideas than to spur concrete legislation. Much of what is proposed would require months, if not years, of research and iterations before arriving at a law or rule.

The AI ​​industry is evolving faster than the rest of the technology sector, which means it is outpacing the federal government by orders of magnitude. Although the priorities listed above are mostly conservative, it is questionable how many of them will remain relevant by the time Congress or the White House actually takes action.