Why AI Probably Won’t Cure Most Diseases in Ten Years: The Verifier Problem
ziv_ravid · x · 2026-08-27
This article questions whether AI can transform biology and drug discovery as rapidly as coding, specifically countering Dario Amodei's prediction that powerful AI could cure most diseases within 5-10 years. The core argument is the "verifier problem": fields like coding and math have cheap, fast feedback loops (compilers, tests), while biological experiments are often slow, expensive, noisy, and incomplete. Drug development requires solving multiple issues, such as correct mechanism and avoiding side effects (e.g., Vioxx), meaning generating candidate molecules via AI cannot overcome the bottleneck of physical-world verification.
More from AGI Musings
- NYT Opinion: The AI Debt Binge Is Endangering the Economy — GaryMarcus · 2026-08-27
- Reid Hoffman: Ask AI Agents for Decisions, Not Updates — reidhoffman · 2026-08-27
- Von Neumann probes possible within 15 years — herbiebradley · 2026-08-27
- Opinion: The biggest risk is moving too slowly, not too fast — Dr_Singularity · 2026-08-27
- Are we prepared for AI to exploit conflicts of meaning instead of resolving them? — PierceLilholt · 2026-08-27
- Non-Tech Users Can Build Useful Internal Apps Using English — tlakomy · 2026-08-27