A long-running Senate task force has released policy recommendations on federal funding for AI. That’s $32 billion a year, covering everything from infrastructure to grand challenges to national security risk assessments.
Although this “roadmap” is not a bill or detailed policy proposal, it nevertheless gives a sense of the scale that legislators and “stakeholders” are always looking at as they grapple with the actual content. However, that is unlikely to happen in an election year. Small enough to disappear.
In a final report released by Senate Majority Leader Chuck Schumer’s (D-N.Y.) office, a bipartisan working group found that Key investment areas have been identified.
Here are some of the main items on the roadmap.
- A “cross-government AI research and development effort that includes related infrastructure” involves DOE, NSF, NIST, NASA, Commerce, and six other agencies and departments to format and share data in an AI-friendly manner. means. In some ways, this seemingly relatively simple task is the most difficult of all, and will likely take years to accomplish.
- Both through the CHIPS Act and others, we will fund U.S. AI hardware and software to work at the semiconductor and architecture level.
- Further fund and expand national AI research resources that have not yet been established.
- The AI Grand Challenge aims to innovate through competition in “applications of AI that fundamentally transform scientific, engineering, and medical processes, and fundamental themes in the design of safe and efficient software and hardware.” It is intended to promote.
- To “support AI responsiveness and cybersecurity” in elections, specifically to “mitigate AI-generated content that is objectively false while protecting First Amendment rights.” It’s probably harder than you think!
- “Modernize the federal government by updating IT infrastructure to leverage the latest data science and AI technologies, and introducing new technologies to find inefficiencies in U.S. statutes, federal regulations, and procurement programs.” Improving the delivery of government services.” I understand what they’re saying here, but this is very frustrating for AI programs.
- Vague but extensive defense-related content, such as “Chemical, Biological, Radiological, and Nuclear (CBRN) AI threat assessment and mitigation by the Department of Defense, Department of Homeland Security (DHS), DOE, and other relevant agencies.” There are many.
- Explore the ‘regulatory gap’ in finance and housing. There, AI-driven processes could be used to further marginalize vulnerable groups.
- “Consider whether other potential uses of AI should be severely restricted or prohibited.” After a section on potentially harmful ones like AI-driven social scores.
- Laws banning AI-generated child sexual abuse material and other non-consensual images and media.
- Ensure that NIH, HHS, and FDA have the tools they need to evaluate AI tools for healthcare and medical applications.
- “Establish a consistent approach to public transparency requirements for AI systems (public and private).”
- Improves “content provenance information”, or the general availability of training data. What did you use to create the model? Is it being used to further train the model? And so on. AI manufacturers will continue to fight this tooth and nail until they can sufficiently sanitize the hoard of ill-gotten data used to create today’s AI.
- Let’s take a look at the risks and benefits of using private AI vs. open source AI (provided the latter exists in an extensible form).
You can read the full report here; There are many other bullet points from which the above (a longer list than I expected to write) came from. No budget numbers have been proposed.
Given that the next six months will be largely given over to election-related fraud, this document will likely serve more to instill interest in a number of popular ideas than to spur actual legislation. . Many of the proposals require months, if not years, of research and iteration before reaching legislation or regulation.
The AI industry is advancing faster than any other technology sector, which means it is outpacing the federal government by orders of magnitude. While most of the priorities listed above are sensible, one may wonder how many of them will remain relevant before Congress or the White House actually takes action.