The White House wants more AI research for less money

The autonomous ship “Sea Hunter”, developed by DARPA in 2016.
The autonomous ship “Sea Hunter”, developed by DARPA in 2016.
We may earn a commission from links on this page.

The White House released a budget proposal this week that at first glance, looks like a big win for the fields of artificial intelligence and machine learning.

The budget for fiscal year 2021 (which begins in October) would ramp up spending for AI research at DARPA (the Pentagon’s research arm) and the National Science Foundation by roughly $549 million. The budget request, which still needs to be approved by Congress, increases AI funding from $50 million to $249 million at DARPA, and from $500 million to $850 million at NSF. A DARPA spokesperson confirmed to Quartz that under the White House proposal, the agency would invest $459 million in funding for “Third Wave” AI research.

But while technologists applaud the increased investment in AI, the White House budget proposal is giving many in the science community pause. Overall, the budget proposes $142.2 billion in spending for research and development, a 9% cut from current levels. In other words, the White House isn’t boosting AI spending as much as asking federal agencies to spend an already shrunken budget on more AI research. 

All the agencies tasked with carrying out the Trump administration’s ambitious plans for AI would be forced to do so with decreased overall funding. Under the new request, NSF faces a cut of $424 million, or nearly 6%. Pentagon research and defense spending would decline under the request from $64.5 billion to $59 billion. NIH, which is also being asked to prioritize AI, will have to do so with $2.9 billion fewer dollars. The Department of Energy’s research and development division faces a cut of $2.2 billion. 

“The President’s budget is very disappointing from the broader scientific perspective. While AI is an important factor for US innovation, underfunding of US basic research hurts US innovation, global competitiveness and economic growth. Thankfully, the US Congress has consistently recognized, on a bi-partisan basis, the importance of basic research and made increases to federal agency budgets year over year,” said Glenn Ruskin, Vice President of External Affairs & Communications at the American Chemical Society, in a statement to Quartz. 

Less research funding for disciplines outside of AI may be a loss for all. The field of AI borrows heavily from other fields, such as mathematics, engineering, and computer science. Martijn Rasser, a senior fellow at the Center for New American Security, found it “disappointing and concerning” that funding for basic research would be down. “We just don’t know where the next breakthroughs will come from,” said Rasser in an interview with Wired.

It’s no surprise that the White House wants science agencies to prioritize AI and machine learning. Both fields are vital to national security and military intelligence, and have amped up their lobbying presence in Washington in recent years. President Trump last year signed an executive order that would make AI a national priority, responding to concerns that China was outpacing the US in AI development. While some estimates suggest that the US has fallen behind China in civilian AI spending, nothing is certain given the lack of precise data from the Chinese government. 

Since 2017, Congress has spared most science agencies from deep cuts proposed by the Trump administration. At the eleventh hour, lawmakers have increased budgets for major science agencies or at least funded them at current levels. But as Science notes, a 2019 budget deal puts caps on both civilian and defense spending, which will give Congress very little wiggle room with federal science funding.

Update: This post has been updated to include comment from DARPA.