NSF launches new AI initiatives for astronomy

The National Science Foundation is launching two new artificial intelligence programs geared towards developing new algorithmic capabilities to advance research in astronomical sciences.

Two new AI institutions, funded in part by the NSF as well as the Simons Foundation, will work with researchers in academia to develop novel AI software tailored to processing both large volumes of astronomical data and images from telescopes that standard AI softwares have trouble computing.

“The AI tools that have been developed by industry are not tailored to our problems,” Andreas Berlind, the program director within the NSF’s Astronomical Sciences Division told Nextgov/FCW. “We don't just have images, we also have text and we have code, and we have spectra, which is another kind of data set, and there aren't models that have been built to deal with those things.”

Although the NSF’s National Artificial Intelligence Research Institutes was launched as an interdisciplinary initiative earlier this year, this marks the first time two institutes will be dedicated towards astronomy research fields. Berlind said that AI software’s capabilities to handle large amounts of data can benefit a multitude of fundamental questions researchers have, such as studying and understanding dark matter and the origin of life in the solar system.

“As with all AI institutes, the goal is to harness these new AI technologies and advance them, and at the same time, impact this other field,” he said. “So it's really timely, given all this data that's coming in, to invest in AI.”

The first institution, called the NSF-Simons AI Institute for Cosmic Origins, will be helmed by the University of Texas at Austin in collaboration with NSF NOIRLab, the NSF National Radio

Astronomy Observatory, the University of Utah, the University of Virginia and UCLA. Its research aims will work to simulate phenomena like the chemical processes within stars. 

The second, the NSF-Simons AI Institute for the Sky, features a partnership with The University of Chicago, the University of Illinois Urbana-Champaign, the University of Illinois Chicago, and the Adler Planetarium. It will work on astrophysics problems such as neutron star and black hole physics as well as the formation of galaxies. 

Researchers are looking to develop AI algorithms suited for astronomical research, including a large language model that is trained on the existing research, as well as algorithms that can read images and other datasets to make connections across multiple data types.

“The algorithms that we need don't exist,” Berlind said. “So these institutes are going to be the trailblazers to doing this work.”