FutureHouse Unveils 'Bio Millennium Problems': The Last Real Eval for AI in Biology
burny_tech · x · 2026-09-21
Edison Scientific and FutureHouse have assembled a set of Millennium Problems for Biology: chosen to be extremely hard to solve yet easy to validate in a simple lab environment. Solving any of them would mark a major advance in biotechnology, and most would materially contribute to curing disease. The authors call this, in a sense, the "last reasonable eval" for AI in biology. Primary work by @MichaelaThinks and @SGRodriques.
Quoting user @srikosuri added a half-joking wish list: mango-banana hybrids, plant-machine interfaces, solution-phase self-assembly of 3D memory chips, a brain the size of a building, browsing the web by thought via BMI, dinosaurs, and more.
Related event: FutureHouse Unveils 'Bio Millennium Problems' as Ultimate AI Benchmark(4 posts)→
More from Research
- Evolution Strategies Beat GRPO on Reasoning Coverage, and Scale Better — burny_tech · 2026-09-21
- Tsinghua's EMERGE-Policy dispatches role-based sub-agents to combine VLA, world models and verifiers — jiqizhixin · 2026-09-21
- Cua-S1 Form Model: Repo, Weights and Dataset Links Released — aigclink · 2026-09-21
- Cua Open-Sources Jev-Style 706K-Param Form Model, Beats Jev API 99.7% vs 83.6% — aigclink · 2026-09-21
- Laya: Open-Weight System 1 Decision Model Answers in 33ms per Pass — Balance- · 2026-09-21
- GPT-6 Astra claims Terminal-Bench Science lead at 65.7%, 31.4 points clear of second place — DeryaTR_ · 2026-09-21