Although competitive StarCraft is in decline, Blizzard is still trying to breathe some new life into its decades old RTS series. During the company’s keynote presentation yesterday at BlizzCon 2016, Blizzard’s president, Michael Morhaime announced, announced new co-op missions and a new co-op commander would be added to the game on November 22. More unexpectedly, he also said that the company would be partnering with Google’s DeepMind project to make StarCraft II a new frontier of competitive gaming AI research.
According to a report by Polygon,
“Researchers interested in using the RTS game to test how AI responds to it will be able to do so early next year. Blizzard is working on modifications for the game that will allow researchers to build systems specifically for the purpose of learning to play StarCraft 2. Those modifications are expected to be ready for release sometime within the first quarter.”
Last year, DeepMind became the first AI ever to beat a professional Go player after it took a series 5-0 against Fan Hui last year. Earlier this spring, the program went on to convincingly beat the best player in the world four games to one.
Scientists, programmers, and researchers have been pitting AIs against pro players in classic games like chess for years, but interest in competitive gaming is more recent. While people have been engineering unbeatable StarCraft AIs for a few years now, applying the resources available to DeepMind against the more complicated sequel is taking the whole endeavor one step further.
In an interview earlier this year with Rock, Paper, Shotgun, Martin Rooijackers, a student who programs AIs for university hosted tournaments, explained some of the challenges involved,
“It may surprise some, but despite being able to perform more actions per minute than a human player, the bots still have sub-par micro-management. Bots have faster individual unit control which allows them to use hit-and-run techniques [dealing damage then moving out of harm’s way], but deciding where/when/how to attack/retreat is still a problem.”
In some ways, the very fact that an AI can do more than a human ever could is also the biggest hurdle it needs to overcome in a game like StarCraft. Unlike Chess or Go, where an AI is analyzing discrete turns and probabilities, things are always changing in a real time strategy game and the fog of war makes means it will need to be able to make assumptions with imperfect information and act on them just like a human would.
As the executive producer on StarCraft II, Chris Sigaty, told The Guardian, the hope is for the project to both make advances in AI research as well as improve the game. “Is there a world where an AI can be more sophisticated, and maybe even tailored to the player?” Sigaty asked. “Can we do coaching for an individual, based on how we teach the AI? There’s a lot of speculation on our side about what this will mean, but we’re sure it will help improve the game.”