There are four core types: rule-based scheduling (predefined dispatching logic with no learning), mathematical optimization such as APS (algorithms like Mixed Integer Programming or Genetic Algorithms find the best sequence within defined constraints), machine learning on historical data (models trained on past production records to recognize patterns), and reinforcement learning trained in simulation (an AI agent learns to schedule by running millions of scenarios in a virtual factory model). Most commercial tools combine elements of more than one type, which is why asking how a system was trained is more informative than asking whether it uses AI.
