AutoDock employs a genetic algorithm to explore the binding modes of ligands to their target proteins. The algorithm iteratively evaluates the binding energy and optimizes the ligand's position, orientation, and conformation. The software uses a scoring function based on a Lamarckian Genetic Algorithm (LGA) to predict the best binding affinity, providing insights into the most stable docking configurations.