Google DeepMind brings AI Co-Scientist from simulation to real-world experiments
omarsar0 · x · 2026-08-28
Google DeepMind released new research taking AI Co-Scientist from simulation to real-world experiments. In CS, it discovered an inference-time scaling architecture beating six frontier models in blinded physician reviews. In chemistry, it designed safe precursor routes for MXenes and drove a semi-automated CVD reactor to produce 2D materials; it also tailored growth recipes to lab constraints in minutes, enabling single-attempt growth of monolayer MoS2, MoSe2, and WS2. In biology, it predicted E. coli swarming phenotypes from sparse data, matching unpublished measurements. 30 domain experts wrote 450 reviews on the generated papers.
Related event: DeepMind's AI Co-Scientist Proves Itself in Real-World Experiments(2 posts)→
More from Research
- Fieldwork data is messy: AI challenge of unstructured reconstruction — anthara_ai · 2026-08-28
- Eval anti-cheat idea: serve models a stale HF cache from before grader fixes — willcb · 2026-08-28
- Study of 400k Claude Code sessions: Domain expertise beats coding skills — alex_verem · 2026-08-28
- Google tests double-blind AI benchmarks with cryptographic protection — The Decoder · 2026-08-28
- LLM-as-judge stuck on one confidence value? Switch to labels. — pedrinho_k2 · 2026-08-28
- Research finds TTT is secretly Linear Attention, speeds up 4x — jm_alexia · 2026-08-28