DeepMind’s next challenge? StarCraft II
Not content with mastering Go and poker, Google’s AI has turned its considerable attention to the world of StarCraft II. Google DeepMind announced its collaboration with video-game developer Blizzard Entertainment on Friday at BlizzCon 2016 in Anaheim, California.
Blizzard Entertainment will make its video game StarCraft II available to AI and machine learning researchers worldwide.
The StarCraft saga is often considered one of the best video-game series of all times, particularly in terms of 1v1 competitive video games. Professional players have been competing in its original version since the late 1990s.
DeepMind is a division of Google that wishes to “push the boundaries of AI” using machine learning and algorithms that help machines solve complicated problems autonomously.
StarCraft will bring tests for Google’s AI to a whole new level, after DeepMind already taught it to compete in games such as Go and Chess, winning over other AIs and human champions.
In a blog post about the new collaboration, DeepMind research scientist Oriol Vinyals wrote: “Over the past five years we’ve helped to pioneer the use of games as AI research environments to drive our machine learning and reinforcement learning research forwards, from 2D games in Atari, to full 3D environments such as Torcs, mastering the game of Go, or our forthcoming DeepMind Labyrinth.”
StarCraft will be a new testing environment for Google’s AI, and DeepMind sees it as a good opportunity to try algorithms that could later be applied in real-world scenarios.
In order to master the game, the machine will have to develop skills including memory, planning over a long time, and adapting its plans according to new information.
Vinyals added: “We’re really excited to see where our collaboration with Blizzard will take us. While we’re still a long way from being able to challenge a professional human player at the game of StarCraft II, we hope that the work we have done with Blizzard will serve as a useful testing platform for the wider AI research community.”