By Matt Hourihan
Congress is expressing growing interest in harnessing the power of artificial intelligence to boost scientific discovery – and AAU recently submitted a set of eight recommendations in response to a bipartisan Senate request for ways to accelerate the pace of American advancement in AI. The recommendations include deeper, broader, and more sustained investments in AI research and AI infrastructure as well as more strategic planning and coordination.
Members of Congress increasingly recognize the significant potential for AI tools to revolutionize fields such as materials science, astronomy, biotechnology, drug discovery, crop science, and others. Harnessing AI as a tool in conducting research and development also offers significant opportunities for economic growth.
In recognition of this potential, last year, two bipartisan House and Senate task forces issued roadmaps for AI policy that would help keep the United States at the forefront of AI innovation and science. The White House is currently working on a new AI Action Plan to achieve a similar goal. Business and scientific leaders have also made the case for investment in AI-enabled science, especially as competitor nations, particularly China, are rapidly advancing their own AI capabilities.
With that as context, Sens. Mike Rounds (R-SD) and Martin Heinrich (D-NM) issued a welcome call for policy ideas to rapidly accelerate the pace of scientific discovery as part of their American Science Acceleration Project (ASAP). AAU has endorsed ASAP and provided the senators with the following recommendations:
- Sustain – rather than cut – America’s science base. Slashing funding for the American scientific research enterprise – as the Trump administration is currently proposing – is not the way to build capacity for AI-enabled science. A better bet would be to maintain strong support for scientific expertise in a broad variety of disciplines, from astronomy to zoology. For years to come, scientists and engineers will still be the ones building and refining AI tools for research and incorporating them into discovery workflows. Science has been the foundation for much of the technological and economic progress of the past century, and now is a terrible time to weaken it. Fortunately, Congress still holds the power of the purse.
- Fund basic and applied AI research. Congress should fund both basic AI research (which looks into the principles, theories, and novel paradigms behind AI and seeks to make it safer, more reliable, and explainable) and applied AI research (which seeks to build effective, scalable AI tools, models, and agents). A robust research program will tackle grand scientific challenges, similar to what Sens. Rounds, Heinrich, and Cory Booker (D-NJ) previously proposed in last year’s AI Grand Challenges Act.
- Support the National AI Research Resource. Congress should authorize and fully fund the National AI Research Resource (NAIRR), a National Science Foundation initiative that connects U.S. researchers with advanced supercomputing resources, testing platforms, datasets, software, and training. There is a broad national consensus that the NAIRR, currently in its pilot phase, is critically important for ramping up AI-enabled science. In the grand scheme of things, the annual funding of $440 million recommended by the NAIRR task force is a wise investment that will reap valuable dividends for the American people.
- Sustain broader cyberinfrastructure investments. NAIRR does not exist in a vacuum. Important NAIRR partners include computing centers at universities like Purdue and the University of Texas at Austin and national labs like Oak Ridge. These partners are important centers of discovery in their own right. As such, Congress should continue to support public supercomputing infrastructure and platforms via the National Science Foundation, the National Institutes of Health, the Department of Energy, and other science agencies. Congress could also consider establishing a research initiative with new funding for energy efficient computing and data centers.
- Get serious on autonomous labs. AI-powered laboratories that will allow scientists to design and conduct experiments remotely, efficiently, and at scale are a source of great promise as engines of scientific discovery. They are also an important front in the global technology race. The United States should aspire to become home to a network of such labs, and Congress should direct agencies to establish a public-private-academic consortium with new funding that will help advance our ability to create and scale up these AI-empowered research tools.
- Data, data, data. There is much the nation must do to build out a secure AI-ready data ecosystem. One important activity is the National Secure Data Service, intended to dramatically streamline access to federal data resources while preserving privacy protections. Congress should work with the Trump administration to ensure the project continues and is adequately resourced. Congress should also ensure that the federal Chief Data Officers Council is empowered to update the Federal Data Strategy and modernize the federal data system so it’s AI-ready. The $150 million funding in the recent reconciliation package for data readiness at the Department of Energy is a good start.
Foster a community of practice in science metrics. As part of the ASAP initiative, Sens. Rounds and Heinrich rightly point out that we need metrics to track scientific productivity. There is currently a robust community of scholars (many on AAU campuses) and federal offices – including NSF, the Government Accountability Office, the U.S. Patent and Trademark Office, and others – that have relevant interest and expertise on this.
However, much of this work is now threatened by funding and workforce reductions – which is why Congress should find ways to actively foster a community of practice in the area of science metrics and measurement. At the bare minimum, Congress should defend funding for offices with relevant expertise. Proactively, Congress should gather input from relevant experts, assess needs, and encourage agencies to develop strategies for improving the measurement of science. Congress could also consider creating a standing interagency body to support the “science of science.”
- Be strategic. The research and computation landscape is increasingly complex, and Congress should have a coherent roadmap to guide public investments. The National AI Advisory Committee has previously recommended an assessment of the current state of cyberinfrastructure and other investments for AI-enabled science to get a handle on needs and gaps. Congress should consider taking up this recommendation and asking for an assessment. It could task the Office of Science and Technology Policy to pursue this through the National AI Office, working with a task force of leaders from industry and science.
Matt Hourihan is associate vice president for government relations at AAU.