The Trump administration is halting funding for a $20 million artificial intelligence institute designed to improve weather forecasting, a move that could sever a pipeline of scientists from receiving training while also hampering the country’s ability to assess the impacts of hurricanes and other weather disasters.
Amy McGovern, the director of the AI Institute for Research on Trustworthy AI in Weather, Climate, and Coastal Oceanography (AI2ES), said the National Science Foundation (NSF) informed the institute last month that it would not renew its five-year grant.
“It destroys it. There’s no other good way to say it,” said McGovern, a professor in meteorology and computer science at the University of Oklahoma.
McGovern said the institute could close next year if it can’t find private funding.
AI2ES is a collaboration among several universities to incorporate AI into weather forecasting and evaluate its trustworthiness.
The move to close AI2ES comes as the Trump administration is heavily investing in AI and accelerating data center construction. The administration’s own AI plan calls for the NSF to develop programs to evaluate AI systems and to build AI job training programs and more AI labs specializing in a range of scientific fields.
The administration in July announced a sweeping plan to promote “global dominance” in artificial intelligence, prioritizing innovation and adoption — both of which are a focus of AI2ES.
“It’s dissonant,” Alan Gerard, a former director at the National Severe Storms Laboratory of the National Oceanic and Atmospheric Administration who tracks new forecasting technology, said of the cuts.
The White House did not respond to a request for comment.
The institute launched in 2020, under the previous Trump administration, as one of NSF’s AI institutes. It has received about $20 million in total funding over the past five years. Michael England, an NSF spokesman, said the agency applauded the AI institutes for their groundbreaking work.
“The U.S. National Science Foundation remains fully committed to advancing artificial intelligence research through the National AI Research Institutes program, a key part of the Administration’s strategy to strengthen U.S. leadership in transformative AI,” England said in an email, adding that “additional award actions remain possible, subject to appropriations.”
NSF and its partner organizations have provided funding for a network of 29 AI institutes. AI2ES was among five institutes up for renewal this year through NSF. Three of the institutes have been renewed and the fourth remains pending, McGovern said.
The Trump administration has proposed a 55% budget cut for NSF, but Congress has yet to pass a budget. The Senate and House appropriations have split from the Trump administration and proposed smaller cuts to science agencies like NSF.
“I really thought because we were an AI institute we would be OK because it was part of the president’s priority,” McGovern said.
The Trump administration’s AI plan recommended that NSF and other agencies to develop workforce programs to expose K-12 students to AI occupations and also create industry-driven training programs to create AI jobs and bolster the country’s AI workforce.
“They want a stronger AI-trained workforce, right? Which we were doing a tremendous amount of,” McGovern said.
She said that the institute provided funding for about 70 roles each year across several universities and that private AI companies had been “snatching our people left and right” because the center had built a pipeline of talent. Among its other measures of success: The institute’s scientists had published more than 130 academic papers and developed AI tools in use by governments today.
The center helped create an AI tool to predict the kind of weather events that can stun sea turtles near the Port of Corpus Christie in Texas, leaving the animals vulnerable to get run over by ships.
It helped build an application that allows forecasters to “see” inside hurricanes even when polar-orbiting satellites with microwave sensors capable of penetrating clouds are not above the storm. To get around that, the application uses data from geostationary satellites that aren’t designed to penetrate clouds and then simulates the inner structures of a hurricane.
The center was also researching how forecasters assess the trustworthiness of private AI tools, like the ones in development by Google and other companies.
“We have social scientists who are diving into work with our actual end users to understand why they trust the AI and why they don’t trust the AI and what they need to do to improve the AI,” McGovern said.
If the center is shuttered, Gerard said it would not harm current meteorological forecasts, but could stymie innovation and put the country at disadvantage.
“A lot of other countries are making investments in AI-related meteorological research, like China. It sets us up to fall behind a lot of other countries in an area that does hold a lot of promise to improve weather forecasting,” Gerard said.