DeepMind's Kevin Murphy: Enabling automated scientific discovery with LLMs

GlenBerseth · x · 2026-08-19

In his keynote at RLC 2026, Google DeepMind's Kevin Murphy presented a vision of treating scientific discovery as a sequential decision-making problem. He emphasized that efficient automated research is now possible thanks to the strength of LLMs as proposal distributions, but noted that model simplicity is also critical for success.

Original post →

More from AGI Musings

AGI Musings channel →