White House official leading the Biden administration artificial intelligence’s agenda is calling for the federal government to increase its research and development spending into this emerging technology.

Arati Prabhakar, director of the White House’s Office of Science and Technology Policy, said federal R&D spending “has continued to grow over time, but it has not kept up with the pace of the economy in the broader innovation system.”

The federal government spends about $200 billion a year on total R&D. Prabhakar said a “pretty modest” share of that spending, about $3-4 billion, is spent on AI research.

“It’s been at that level, growing modestly for quite a while. So, we’re not seeing a huge surge yet in public spending, partly because budgets are constrained,” she said Tuesday at the Brookings Institution.

The federal government is getting repeated calls to ramp up its AI spending, citing national security concerns and a technological arms race with nation-state adversaries.

The National Security Commission on AI, a blue-ribbon panel commissioned by Congress, recommended in its final report that the federal government exponentially increase its AI R&D spending to $32 billion annually by 2026. 

“That is exactly the direction that we need to move,” Prabhakar said.  “It needs to be a sizable percentage, if we’re really going to seize this moment.”

Michael Kratsios, the U.S. Chief Technology Officer under the Trump administration, also called on the federal government to nearly double its level of AI research spending in 2019.

The U.S., accounting for both the public and private sectors, spends about $800 billion a year in total R&D — about 3.5% of the country’s gross domestic product.

Erwin Gianchandani, assistant director of the National Science Foundation’s Directorate for Technology Innovation and Partnerships, raised concerns about the government’s “under-investment” in AI, at a time when competitors like China are pouring billions of dollars into AI breakthroughs.

Meanwhile, Gianchandani said Congress shrank NSF’s overall budget this year by about 5-8%, compared to last year’s funding levels.

“The budgetary shortfalls are incredibly problematic. Every year at NSF, for instance, it’s a couple billion dollars’ worth of projects, that are about as competitive as the projects that we are able to fund, that we just can’t get to because of appropriation shortfalls,” he said. “That’s the budget shortfall, and that is lost opportunity.”

Prabhakar called China’s growth in R&D spending “unparalleled,” but said U.S. private-sector research spending in AI has kept the country from being outpaced by its adversaries.

“Our private sector has really put the pedal to the metal on R&D spending. It’s the intensification of the innovation economy. It’s where our AI advances have come from, but many more things, especially in the IT sector,” she said.

Prabhakar said the federal government needs to keep providing steady funding for AI research — especially for long-term research the private sector might deem too risky for investment.

“We’re in the hype cycle, and so some of the noise is going to get washed out. And what that looks like is there will be some places where people will retreat on R&D spend. So, all of that is going to happen in the private sector. But to me, it underscores that it’s only the federal government, it’s only we as a public that are going to support the research that is long-term in nature. That’s the deep foundational research that reaches across every part of the country, not just where people have business operations,” Prabhakar said.

NSF Director Sethuraman Panchanathan said last month that today’s AI breakthroughs were made possible by about 60 years of federally funded research.

The Biden administration, under a sweeping AI executive order, is also ramping up the use of AI tools across the federal government and staffing up with AI experts.

“Last year really became a moment that the president and the vice president seized to say, first of all, this is the most consequential technology of our times,” Prabhakar said.

“At the end of last year, I looked at the enormous progress that this administration had made in getting us on the right track. There’s a lot more to do, but we got aimed and started in the right direction on managing risks — and you have to do that. But the reason you’re doing that is to build a stable platform so that you can reach for the stars,” she added.

At an “AI Aspirations” summit last month, OSTP showcased how a dozen federal agencies are up to the task of using AI to deliver more services to more people.

“It’s about putting AI to work to achieve the big things that this country needs to do, from health to dealing with climate change, to lifting up opportunity for every person,” Prabhakar said.

Among agencies’ aspirational projects, she said ARPA-H is working on a program with the ultimate goal of the Food and Drug Administration using AI to accelerate the approval of new medications to months, rather than years.

Surveys conducted last year by the Artificial Intelligence Policy Institute think tank show most respondents remain wary of AI, and most support the Biden administration’s push for stronger AI regulation in an executive order last fall.

“Many people in this country don’t trust AI and some of the polling still shows, I would say, somewhat more fear than hope in the United States. And until that trust is there, we’re not going to realize its business potential. We’re not going to realize its potential for public missions as well,” Prabhakar said.

“They’ve charted the right course, we’re doing a lot through executive action through how the government uses AI. And I think that’s exactly the right start,” she added about the administration’s AI policies. “There’s so much more to be done, including legislation. But I do think that that’s on the right track. I’ll also tell you, ironically, when you get to the seizing of benefits, it’s in building AI systems and putting AI to work, that you actually figure out what the real risks are. And you find the path to build in the mitigation.”

To ensure the U.S. keeps producing more AI experts, the NSF recently launched its Technology Innovation and Partnerships Directorate, which is focused on growing the STEM workforce.

The agency is also leading the National AI Research Resource, a hub for the federal government to share its AI and data resources with trusted academics and private-sector researchers.

The NAIRR is currently a pilot project with about $30 million in funding from Congress — far from the $2.6 billion that the task force behind the project requested for the next six years.

“It is, candidly, a drop in the bucket, relative to what the task force articulated was the need here,” Gianchandani said.

“We instituted the CHIPS and Science Act to try to course-correct. My big worry is five years from now, 10 years from now, at some point in the future — maybe it’s even sooner than that — we wake up one morning, and we have those same issues not just in semiconductors and chips, but in AI and in quantum and in biotechnology, and so forth. I think that’s really what’s on the line here,” he added.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.