Scoring only works on molecules that already exist in a collection. AI can also propose new ones: given the target and the properties that matter, generative models build structures that never existed before, then refine them step by step toward better predicted potency and drug-like behavior. The output is a shortlist of designed candidates, still hypothetical until someone makes and tests them.
Can AI Run a Drug Discovery Lab on Its Own?
Where AI Already Helps Today
Making Molecules, Not Just Judging Them
2 / 4
Filtering still assumes the right molecule is already sitting in the collection. Often it is not, and that is where the second role comes in: AI that proposes molecules instead of judging them.
0:00 / 0:00