The Next AlphaFold Won't Design Drugs — It Will Find the Right Targets
julia_kiseleva · x · 2026-08-19
Julia Kiseleva argues that AI drug discovery is overly focused on molecule design, while many programs actually fail earlier at target selection. The next AlphaFold-scale breakthrough will come upstream: treating genomes as a compressed record of billions of years of evolutionary experiments, and building foundation models that read latent phenotype from the language of evolution to discover the right targets.
More from AGI Musings
- Nvidia Bets on Open-Source AI Explosion — peterwildeford · 2026-08-19
- Parts of the AI world are built on shaky conceptual scaffolds — sebkrier · 2026-08-19
- Expert Rants: Don't Replace Keyword Search with AI — generativist · 2026-08-19
- Desire for 'Revealed Mind' Explains Aversion to AI Art, New Study Suggests — RobbWiller · 2026-08-19
- Opinion: We are lucky the AI race leaders are Sam, Dario, and Demis — AdrienLE · 2026-08-19
- AI boosts homework scores by 18% but lowers exam results by 20% — luisdans · 2026-08-19