DeepMind on agentic AI as a “conjecture machine” for science
JMateosGarcia · x · 2026-07-20
Google DeepMind’s essay frames agentic AI as a “conjecture machine” for science: it can generate hypotheses and candidate solutions cheaply and at scale, but the bottleneck shifts to validation.
The piece argues that refutation remains slow and expensive because real-world testing still depends on wet labs, peer review, and other physical or institutional processes. So AI may flood science with ideas, while the rate-limiter becomes how fast those ideas can be checked against reality.
More from AGI Musings
- New paper: AI agents negotiate badly under uncertainty and break deals on purpose — menhguin · 2026-10-26
- NBER: Only 2.9% of US consumers pay for AI; retention as low as 20%, pricing power at risk — avicgoldfarb · 2026-10-26
- AI researchers guessed robots wouldn't outrun humans until 2034 — it already happened — ben_j_todd · 2026-10-26
- AI folks, stop saying "biology" when you mean drug development — KevinKaichuang · 2026-10-26
- ESR on the Bugpocalypse: locking down frontier models only hands the game to Chinese open weights — colinmcnamara · 2026-10-26
- Anthropic: junior tech hires fell from 27% to 9% as agent hours hit 6x employee hours — thione · 2026-10-26