AI earns its place where the space of possibilities is too large to search by hand and the evidence is weak and scattered. In this field that means reading DNA at scale, combining many small genetic signals into a ranked list of candidate targets, and filtering candidate molecules before anything is made.
The pattern behind the three
Each stronghold is a search problem, not a knowledge problem. The model is not telling scientists something new about biology; it is ordering a field that is too wide to examine item by item. That is why the value shows up as concentration — the same laboratory budget covers more ground because unpromising options are set aside early.
Why scale alone forces the hand
A single human genome is about three billion letters. A study comparing thousands of people therefore involves trillions of letters. Reading that by eye is not slow — it is impossible. The task is not hard because the biology is subtle in every position; it is hard because there is so much of it.
It is tempting to say AI helps "everywhere" in drug discovery. It does not. It helps most where the search is wide and the signals are faint. Where a single decisive measurement settles the question, a laboratory assay is still the better tool.