Insights from BTD26: Daniel Grün on AI in Cosmology

How many person-years does it take to understand the universe?

Without artificial intelligence, future cosmological experiments will no longer be feasible. At our recent Big Techday 26, Daniel Grün from LMU Munich explained why astrophysics has reached the limits of human analysis capacity and now depends on AI methods.

Daniel demonstrated how generative AI models create simulations of the universe in seconds, producing results that are statistically indistinguishable from original simulations that require millions of CPU hours. He showed that neural networks determine physical parameters from gravitational lensing effects three times more precisely than conventional statistical methods. He also highlighted that future analyses without AI automation would require thousands of person-years of manual work.

Watch the full recording here.

Thanks to Daniel for sharing his insights on how AI supports cosmology research and opens up new possibilities.