Inaugural Cosmic Horizons Conference Unites Astronomers and AI Experts
Brian Scott Mason, staff scientist at the National Radio Astronomy Observatory
In early May, NSF-Simons AI Institute for Cosmic Origins (CosmicAI) brought together leading voices in astronomy and artificial intelligence (AI) for the 2025 Cosmic Horizons Conference, a three-day event held May 6-8, 2025, at the University of Texas at Austin. The conference explored the rapidly evolving intersection of artificial intelligence, machine learning, and astronomy research. Conference sponsors included the National Science Foundation (NSF), Simons Foundation, and the Oden Institute for Computational Engineering and Sciences.
Astronomers and AI researchers presented their work to a broad range of attendees, which included faculty, students and industry professionals from across 40 different institutions. More than 40 presentations provided a platform for interdisciplinary dialogue, learning from one another’s methods, challenges, and perspectives across both domains.
As Alessandra Corsi, W. H. Miller professor in the Department of Physics and Astronomy at Johns Hopkins University, noted during an interview, “We are at a very exciting time – when we are starting to see a trickle of multi-messenger discoveries, but we think that over the next decade, we’re going to see a flood. Astronomers on their own don’t have the tools to really capitalize on this. Talking to people who are experts in data science and AI and bringing people together in a conference like this can really enable us to make the next big leap in multi-messenger astrophysics.”
Alessandra Corsi, W. H. Miller professor in the Department of Physics and Astronomy at Johns Hopkins University
Presenters shared research on using machine learning to study galaxies, dark matter, gravitational waves, and the structure of the cosmos. Several projects focused on building better AI tools, running simulations to speed up discovery, and using AI to make sense of large astronomical datasets. Other presentations focused on how to make AI systems more transparent and trustworthy for scientists. From understanding the life cycle of stars to spotting strange or unexpected patterns in space, the presentations showed the wide range AI is being used in the field of astronomy.
The conference centered on three core themes: advancing scientific discovery through AI, building trustworthy and interpretable AI tools, and preparing for the big-data era in astronomy. Speaking to the importance of transparency and trust, Aimee Lee Schechter, PhD Candidate at the University of Colorado Boulder, noted, “I think the more the scientific community can show in their work why we trust these models, the more it will help us understand the place that AI can have productively in our society overall.”
Stella Offner, Randi Ludwig, Eric Murphy, Kevin Gullikson, and Tanmoy Laskar
To support the next generation of researchers and leaders, a panel explored career-building strategies, networking and opportunity development. The panel was led by Kevin Gullikson, Machine Learning Researcher at Trase, Randi Ludwig, professor at UT Austin, Tanmoy Laskar, assistant professor of Physics and Astronomy at the University of Utah, and Eric Murphy, Astronomer at the National Radio Astronomy Observatory (NRAO). During the panel, they expanded on the pros and cons of various professional pathways, and gave personal insights and experiences to help emerging scientists navigate in rapidly evolving fields.
Murphy underscored the unifying power of astronomy, saying, “What astronomers do in general, by studying our place in the universe, is important to humanity because it is something that really links all of us - independent of your culture, religion, or what your day-to-day life is like. We all live on the same planet, and we’re all interested to understand how we got here and why we are here.” CosmicAI aims to host an Astro-AI meeting every year of its 5-year mission.
A full list of conference speakers and panelists can be viewed here.