Researchers and companies have begun using artificial-intelligence systems to generate antibody sequences with drug

Researchers and companies have begun using artificial-intelligence systems to generate antibody sequences with drug-like properties. The shift is a design-stage development: software can propose candidates that scientists then refine and test in the laboratory.

The important limit is that a plausible antibody is not yet a medicine. Researchers still need to show that a candidate binds as intended, can be manufactured, is safe enough to study, and produces a meaningful result in people.

There is no single agreed definition of an AI-designed antibody, and the field had not yet produced a clear clinical success from one. The next proof is not a model demonstration but independent experimental and human evidence.

Read the full story at STAT ↗