گول این حرفا که «هوش مصنوعی قراره سرطان رو درمان کنه» رو نخور، چون یکی از استادای بزرگ این حوزه دست رو نقطه‌ضعف اصلی گذاشته

A pioneer of AI-in-biology just published a sharp rebuttal to the "AI will cure cancer" narrative. The core argument: It's a measurement problem. Over 90% of drugs that enter clinical trials still fail — a number that hasn't moved in decades. In most cases, the molecule works fine. The mechanism it was built to hit was simply wrong. No amount of AI-designed molecules fixes that: Building better keys for the wrong locks does not open the correct doors. The reason we keep picking the wrong locks: human biology has never been measured at the scale or resolution needed to actually map disease mechanisms. LLMs can connect results from the literature, but they can't extract answers about molecules and processes that were never captured in the first place. This is exactly where scalable mass spectrometry-based proteomics has a role to play. Unlike sequencing, MS can directly quantify the functional molecules — proteins, their modifications, their interactions — that actually execute cellular decisions. As single-cell and low-input proteomics methods mature, they offer a path to the kind of large-scale, perturbation-aware functional datasets Koller argues are missing: not just what's genetically possible, but what's biologically happening. Substantive AI-for-biology progress requires the right instruments pointed at the right layer of biology.