Universities and nonprofits press Congress: Help us keep up on AI
Illustration: Aïda Amer/Axios
The private sector, notably OpenAI, Google and Microsoft, is dominating the generative AI race, and universities and researchers want Congress to help them keep up.
Why it matters: In the United States, private industry is leading the way on advancements in AI, with hefty investments from well-resourced companies, spurred along by the frenzy over generative AI applications like ChatGPT.
- Now some lawmakers are preparing to introduce legislation to fuel AI R&D from members of the public.
- Academics, nonprofits and other researchers fear that AI progress will be motivated only by profit, without the longer timelines, ethical considerations, and checks and balances present in research settings.
- There's also concern that without a more diverse group of researchers working on AI and training datasets, biases and misinformation will proliferate.
Background: The National AI Initiative Act of 2020 established a federal advisory committee called the National AI Research Resource Task Force.
- In January, a report from the NAIRR Task Force recommended that Congress appropriate $2.6 billion over six years to stand up NAIRR for AI research at universities and other institutions, including for students learning about AI and educators using AI tools.
Driving the news: Rep. Anna Eshoo, co-chair of the Congressional AI Caucus along with Rep. Mike McCaul, is looking to introduce a bill to authorize NAIRR, a House staffer told Axios, developing legislation based off the recommendations in the report.
- On the Senate side, Sen. Martin Heinrich, who co-chairs the Artificial Intelligence Caucus along with Sen. Mike Rounds, is working on a bill to authorize NAIRR with an aim for it to be included in future comprehensive AI legislation, a Heinrich spokesperson told Axios.
- A Rounds spokesperson said the senator’s office is "currently reviewing this legislation."
- The National Artificial Intelligence Research Resource Act, as outlined as a draft in NAIRR's report, would stand up NAIRR as an institution and authorize its funding.
- The draft text says it would target $440 million a year from 2023 to 2028.
Details: The draft bill reads: "Access to the computational resources and datasets necessary for artificial intelligence research and development is often limited to very large technology companies and well-resourced universities...
- The lack of access to computational and data resources has resulted in insufficient diversity in the artificial intelligence research and development community."
By the numbers: Microsoft announced in January it was investing $10 billion in OpenAI, kicking off a well-financed race over generative AI development.
- Google invested $300 million in Anthropic in February; the company announced a $450 million funding round this week.
- Per PitchBook data, generative AI startups raised more than $1.6 billion just in the first quarter of 2023. Both Microsoft and Google have stated support for getting nonprofits, universities and researchers funding to participate in training AI.
- "Academic research plays a role in advancing accountability," Microsoft president Brad Smith told Axios at an event Thursday in Washington. "We do need to find a new way to provide broad-scale computational resources to academic researchers in the United States and other countries."
What they're saying: "The key aspect here is, who has a seat at the table?" Russell Wald, director of policy at Stanford University's Institute for Human-Centered Artificial Intelligence, told Axios.
- "Not one university today, not MIT, not Stanford, not Carnegie Mellon, could train a ChatGPT. ... You're going to have to have a lot of money to be able to purchase that level of computing."
- Wald said past tech and science breakthroughs were made possible by government backing and public-private partnerships, such as the Human Genome Project.
Be smart: There are a lot of cooks in the kitchen when it comes to the government trying to regulate AI, with the Biden administration putting out its own roadmaps and guidelines along with leadership in Congress looking to get involved.
- It could get complicated fast, with overlapping ideas, competing goals and Republicans' hesitancy for the government to spend more money.
Yes, but: There is bipartisan interest in the U.S. leading on AI policy and making investments in research, which is seen as a hopeful sign that this kind of federal funding could get tacked onto a spending bill.
The bottom line: As the U.S. figures out its plans for governing AI and making research around it more equitable, the private sector will just keep racing.
