Keeping fans hooked is how game developers make money on many “free” video games. Now, IBM and PlayFab are announcing an AI project to help game makers do just that. The initiative is starting with Atari’s “Roller Coaster Tycoon Touch,” a “Farmville”-like mobile game in which players build their own amusement park.
PlayFab CEO James Gwertzman says AI can supercharge the current practice of studying user data to improve games. Natural language processing can scan through reviews to find common praise or complaints about a game, for instance, and machine learning can find patterns across the game histories of millions of players. One area AI can help in is matchmaking—pairing two people who will enjoy playing (and consequently, spending money)—together online. “One of the things we’re looking forward to is using machine learning to develop better matchmaking algorithms,” says Gwertzman.
As is often the case, IBM is making big claims about Watson’s ability in this area before it has much evidence. IBM’s AI has been chewing on “Roller Coaster Tycoon” user data for about three weeks, but the companies don’t yet have examples of how it has improved the game. IBM and PlayFab are just now inviting other game makers to sign on.