Game Smarts

Artificial intelligence has been a part of videogames since their early days. For game developers, Al has come to mean the broad range of techniques used to generate the behavior of these opponents, battlefield units, team mates, NPCs, or anything else that acts in the game with simulated intelligence. A few of these techniques, such as finite state machines and the heuristic A* search algorithm, have proven themselves in many games over the years. Following the A* search algorithm´s path, game developers are starting to explore techniques from several AI research subfields, including automated planning and machine learning. Machine learning has the potential to let AI characters improve with experience and adapt to individual players. The two machine learning techniques most commonly discussed in the games context are inductive and reinforcement learning