The White House’s Science Strategy Gets Two Big Things Right | American Enterprise Institute
The Trump administration has a contentious relationship with America’s scientists, strong-arming universities and appointing vaccine skeptics to senior roles. A recently released White House report aims to reset the country’s science strategy—and many scientists will naturally be skeptical. Yet its two core ideas are worth taking seriously: harnessing AI to accelerate science and redesigning research organizations for this purpose.
First, the backdrop: science and R&D are more important than ever to economic growth and technological leadership. Yet America’s competitive position has deteriorated. China now spends slightly more on R&D as the US when adjusted for purchasing power. By any metric, China is making enormous strides in catching up to the US in science.
What is to be done? Michael Kratsios, who directs science and technology policy in the White House, has two big ideas. The first is that the government should retool to better apply AI to science. Amid all the claptrap about artificial intelligence, this deserves focus. The greatest scientific advances of the last half century have all depended on huge volumes of computing. Computers were as important as DNA sequencing machines in decoding the human genome. The Higgs boson and gravitational waves both required supercomputers to detect. The world’s greatest physicists cluster around the world’s biggest supercomputers for a reason.
The effort to mobilize AI is particularly important because the US can’t compete on the number of researchers, where China now has a staggering quantitative advantage. Competing on researcher quality is also tough when China’s best universities are world class. The only realistic strategy for staying at the cutting edge is to harness the best possible tools.
Artificial intelligence is already transforming the scientific toolset, as the Nobel committee recognized when it gave Demis Hassabis and John Jumper their 2024 prize in chemistry for inventing AlphaFold. Companies and researchers alike are harnessing new AI tools for materials science and biology. Bo Wang, a University of Toronto professor who specializes in AI and biology, recently noted that graduate students are asking potential academic advisors: “how many GPUs do you have?”
Questions like this suggest it’s also time to rethink how research labs are organized and funded. This is the second takeaway from the White House strategy, which notes that America’s science funding structures date from the post-World War II era. As these government funding arms and the universities that execute research have aged, they move more slowly. A recent high-profile report at Yale University noted that “bureaucratic expansion” had “undermined the university’s academic mission.” Another study found that federally-funded researchers allocate around 45% of their time to administration, including proposal drafting, reporting, and budgeting.
Concern over the slowing of American scientific progress has motivated a new field called metascience, which analyzes how effectively scientific institutions foster and disseminate new ideas. The Institute for Progress, a think tank that’s pioneered this work, ascribes the decline in scientific progress partly to the fact that the U.S. government has become a more conservative and bureaucratic funder of science. The average grant recipient from the National Institutes of Health is over a decade older today than in 1980. Younger researchers and those with higher-risk ideas are crowded out.
The White House’s science strategy draws heavily on these critiques and embraces many of their proposed solutions. One idea is to shake up grantmaking by giving grants to researchers directly rather than to the institutions that house them.
A second is to experiment with new funding architectures beyond universities. Specialized research institutes have a long history of driving scientific progress, like how Cold Spring Harbor Laboratory fostered the field of molecular biology or the Institute for Advanced Study did the same for quantum physics, game theory, and computer architecture.
An early example of what this might look like is the Department of Energy’s Genesis Mission, which links AI companies, national laboratories, and universities to fund research into topics like using robotic laboratories to accelerate research. The first era of computing gave birth to entirely new fields of inquiry, from computational biology to computer science itself. Grad students are already assessing research labs by their GPU count. Government must also prepare for entirely new modes of research as AI transforms science.
