Daphne Koller says AI will help biology, but biology is still too poorly understood
DaphneKoller · x · 2026-08-04
- Daphne Koller argues that AI will eventually transform human health, but rejects the idea that a clever enough reasoner can simply infer cures from existing biological knowledge.
- She notes that only about a quarter of human diseases have an approved therapy, and more than 90% of drugs entering clinical trials still fail.
- In her view, many failures are not because the molecules are poorly engineered, but because the underlying biological mechanism was misunderstood from the start.
- The attached Insitro image frames the essay as a “Deep Phenotype” theme: scaling biology and deep causality, not just scaling model capacity.
Related event: insitro CEO Daphne Koller: AI Drug Discovery Has No Magic Wand Yet(7 posts)→
More from Research
- Reverse-Engineering NVIDIA Blackwell Tensor Cores for Bit-for-Bit Software Simulation — ycombinator · 2026-08-05
- Analyzing the Four Core Paradigms of Modern In-Context TTS — rdesh26 · 2026-08-05
- Graph Analytics Benchmark Graph500 Selected for SPEC CPU 2026 Suite — Prof_DavidBader · 2026-08-05
- NeurIPS 2026 Competition: Predicting Fusion Plasma Equilibrium with ML — wattmaller1 · 2026-08-05
- Researcher Calls for a Refutations & Critiques Track at NeurIPS — RylanSchaeffer · 2026-08-05
- MuScriptor: Open-Weight Multi-Instrument Music Transcription Model Released — drscotthawley · 2026-08-05