Expediting Astronomical Discovery with AI Agents: Progress, Challenges, and Future Directions
by
TDLI, Open Area (4F-SW)
Tsung-Dao Lee Institute
Host: Fabo Feng
Join Tencent Meeting：https://meeting.tencent.com/dm/FjvoVc8KwcfB
Meeting ID: 137369162 (no password）
Abstract:
The expansive, interdisciplinary nature of astronomy, combined with its open-access culture, makes it an ideal testing ground for exploring how Large Language Models (LLMs) can accelerate scientific discovery. Recent developments in LLM reasoning capabilities have shown substantial progress—our work demonstrates that AI agents can now achieve gold medal performance on International Olympiad on Astronomy and Astrophysics (IOAA) problems, indicating their growing analytical abilities. In this talk, I will present our recent advances in applying LLMs as agents to real-world astronomical challenges. We demonstrate how LLM agents can conduct end-to-end research tasks in galaxy spectral fitting, encompassing data analysis, strategy refinement, and outlier detection—approaching capabilities similar to human intuition and domain knowledge. However, limitations remain. While autonomous research agents could theoretically help analyze all observed sources, the cost of closed-source solutions remains prohibitive for large-scale applications involving billions of objects. Additionally, the Moravec paradox manifests clearly in astronomy: tasks requiring abstract reasoning may be easier for AI than seemingly simple perceptual tasks. Current models still struggle with chart reading, multi-modal data interpretation, and other fundamental astronomical workflows. To address the cost limitation, we developed lightweight, open-source specialized models (AstroSage) and evaluated them against carefully curated astronomical benchmarks. Our research shows that these specialized LLMs can outperform larger general-purpose models on astronomy Q&A tasks when appropriately pretrained and fine-tuned, demonstrating a path forward for building more capable and accessible astronomy-specific models. Looking ahead, the path forward involves integrating more function-calling tools and building a comprehensive ecosystem—not just better models. The astronomical community's collaborative infrastructure will be important for scaling up automated inference and expanding the role of AI in astronomical research.
Biography:
Yuan-Sen Ting is an Associate Professor of Astronomy at The Ohio State University. His research applies deep learning techniques to statistical inference in astronomy, drawing on data from large-scale surveys across spectroscopy, astrometry, photometry, and time-series observations. His work spans galactic evolution, stellar spectroscopy, and cosmology, with recent efforts focused on developing large language models as autonomous research agents for scientific discovery. A native of Malaysia, Yuan-Sen holds a PhD in astronomy and astrophysics from Harvard University (2017), funded through a NASA Earth and Space Science Fellowship, and concurrent Bachelor's and Master's degrees from the National University of Singapore and École Polytechnique in France. Following his doctorate, he was awarded the NASA Hubble Fellowship, Carnegie-Princeton Fellowship, and Institute for Advanced Study Fellowship—held jointly across Princeton University, the Carnegie Institution for Science, and the Institute for Advanced Study—before joining the faculty at the Australian National University and subsequently Ohio State.
Yuan-Sen has authored over 240 publications, including work in Nature, Nature Astronomy, and a review on deep learning in astrophysics for the Annual Review of Astronomy and Astrophysics. He has also written Statistical Machine Learning for Astronomy, a textbook providing a systematic treatment of machine learning methods for astronomical research. His honors include the Alexander von Humboldt Fellowship, the Australian Research Council DECRA Fellowship, and recognition as an AURA Future Leader. He currently chairs NASA's AI/ML Science and Technology Interest Group and previously led the NASA Cosmic Origins Program Stars Interest Group. Beyond academia, Yuan-Sen wrote a monthly column for Malaysia's Sin Chew Daily, delivered a TEDx talk on AI and humanity in Kuala Lumpur, and has produced educational content for TED-Ed that has been viewed over four million times.
