Frontiers in Artificial Intelligence for Science, Security, and Technology (FASST) is an initiative within the US Department of Energy.
Premise
Quote
FASST will transform the vast repositories of scientific data produced at DOE user facilities to be AI-ready and build the next-generation of highly energy efficient AI supercomputers. This national AI capability will allow U.S. researchers, including the 40,000 scientists at the national labs, to develop trustworthy foundation AI models to realize breakthroughs in a variety of scientific and energy applications.
FASST has four pillars which are described in the FASST fact sheet (energy.gov):
- AI-ready data: Create a repository of DOE datasets that are “AI-ready” (whatever that means). Classified and unclassified data are both in scope, and these data will be available for government, industry, and the scientific community.
- Frontier-scale AI computing infrastructure and platforms: Next-generation supercomputers, ESnet, and public-private partnerships.
- Safe, secure, and trustworthy AI models and systems: Develop frontier models for science.
- AI applications: Use AI to specifically advance science domains.
Outcomes
The FASST pitch offers the following outcomes:1
- AI models for national security and defense.
- “Attract, train, and retain top scientific talent.”
- Apply AI for scientific discovery.
- “Unlock new clean energy sources” - whatever this means.
- AI governance. This seems like the most compelling aspect.
Schedule
I don’t know what the schedule is because it is not funded. As of November 2024, there is a Notice of Request for Information (RFI) on Frontiers in AI for Science, Security, and Technology .._ out which is due on November 11.