Abstract

Invited Talk - Plenary

Friday, 11 September 2026, 09:30   (MW-1801 / virtual plenum)

AI in Astronomy

Caroline Heneka
Heidelberg University

Astronomy is entering an era of unprecedented data volume and complexity, from galaxy surveys to line intensity mapping and radio maps of the early Universe, outpacing what classical analysis pipelines can extract. Modern AI offers a path forward: generative models that simulate realistic skies in seconds rather than days, self-supervised networks that learn rich representations directly from data, and simulation-based inference that turns these into calibrated, uncertainty-aware constraints on physics. In this talk, I will showcase how these tools are reshaping the astronomer's workflow end-to-end, from fast, flexible simulation to robust inference and foundation models, and highlight what makes them trustworthy for science: resilience to noisy, out-of-domain, and mismatched data. Drawing on examples across galaxy evolution, large-scale structure, and the Epoch of Reionization, I will argue that AI is becoming not just a tool for speed, but a foundation for discovery in the era of instruments such as the Square Kilometre Array and beyond.