Special Session SS10  1-2 July 2024

The impact of the rapidly evolving field of artificial intelligence on astrophysics research: avenues and potential breakthroughs

News: Confirmed Invited Speakers: * Viviana Acquaviva * Noemi Anau Montiel * Luisma Sarro * Ashley Villar * Mike Walmsley

Aims and scope

Over the last decade, machine learning (ML) and artificial intelligence (AI) methods have become widely used across various fields of astrophysics. These techniques have sparked a new era of data-driven discoveries, significantly enhanced our ability to extract model parameters from data, and are increasingly being used to supplement or even replace costly numerical simulations ranging from cosmology to stellar structure and evolution. As these methods become integral to astrophysical research, we're witnessing the emergence of AI for astronomy as a distinct field. This development is driven by the realization that state-of-the-art AI tools are often not ideally suited for our unique data sets or for extracting the physical meanings inherent in astronomical data. Customizing AI to meet the specific needs of astrophysics is becoming essential. The rapid expansion of AI, with more ML methods becoming standard in astrophysics, suggests that novel methods and developments could be central to the next major breakthroughs in our field. The advent of Large Language Models based on transformer architectures is a prime example of how AI innovation can transform astronomical research, when adapted for this domain.

In this special session, we will delve into the prospects of artificial intelligence in astronomy over the next decade. We aim to explore methods and research avenues that are most likely to lead to significant advancements. A particular focus will be on specific astrophysical problems that could benefit from tailored AI solutions. Examples of these methods include...

* Foundation AI models for astrophysics * LLMs and multi-modal learning for astrophysics * Uncertainty quantification in ML models * Physical Anomaly detection * Simulation-Based Inference

We will cover a broad range of topics, from cosmology to exoplanets, and will get a perspective from a diverse cohort of invited and contributed speakers. Rather than focusing on well established methods and solved problems, we ask the speakers to provide a vision of how the rapid evolution of AI methods outside of astronomy can be tailored for the resolution of fundamental problems in astrophysics. This session is organized by AstroAI, a new center based at the Center for Astrophysics | Harvard & Smithsonian that is dedicated to the design and development of artificial intelligence for astrophysics, to enable next generation research.


  • Foundation AI models for astrophysics
  • LLMs and multi-modal learning for astrophysics
  • Uncertainty quantification in ML models
  • Physical anomaly detection
  • Simulation-based inference

Invited speakers

Scientific organisers

Rafael Martinez-Galarza Cecilia Garraffo Marc Huertas-Company Andres Moya Floor Broekgaarden


Rafael Martinez-Galarza jmartine @ cfa.harvard.edu

Updated on Tue Feb 27 14:28:39 CET 2024