AI for protein folding refers to machine-learning systems chiefly deep neural networks that infer a protein’s three-dimensional structure from its amino-acid sequence and related biological data, producing atomic-level models far faster than traditional experimental techniques.