This is an action-packed newsletter about AI in games, in particular from a research perspective, so strap yourselves in! We've included our report from the Computational Intelligence in Games 2014 conference, as well as its collection of white papers to download and read (on top of the usual round-up).
Download the newest AI Sandbox build together with the first simulation we're releasing within The AI Sandbox™ focusing on Monte-Carlo Tree Search (MCTS). This algorithm is revolutionizing search-based AI, in particular board games and now increasingly AAA games. Applying it in practice often requires specialised insights, that's where the AI Sandbox' visualizations prove extremely useful.
It's been an interesting week here in the AiGD secret labs. Over the weekend, the site was taken down briefly by an accidental DoS, which nevertheless took time to fix. On Monday, the electricity was out as one of our R&D machines blew the main office fuse, also taking the day to repair. These will make for great stories after our launch is over!
Monte-Carlo Tree Search is a powerful AI algorithm that was recently discovered and in the past years has found great success in board games. This year, MCTS finally hit the AAA games industry, specifically thanks to Creative Assembly's TOTAL WAR: ROME II. Find out more about the design motivations for using the algorithm, how it was integrated into the game, and tips & tricks used to make it work efficiently.
City building games walk a fine line between AI that's fun to watch and a balanced simulation. The implementation also needs to be very efficient to support acceleration and hundreds of villagers. This interview digs into medieval-themed Banished. Luke Hodorowicz shares insights about the individual AI code, and designing a consistent experience.