Google DeepMind and Blizzard join hands to use StarCraft 2 for AI research

Google's DeepMind is teaming up the Blizzard Entertainment to use real-time strategy game StarCraft 2 to train its artificial intelligence (AI) systems. The London-based AI research firm and subsidiary of Google announced that the gaming company will release an API early next year allowing "researchers and hobbyists around the world to build and train their own AI agents to play StarCraft 2."

"DeepMind is on a scientific mission to push the boundaries of AI, developing programs that can learn to solve any complex problem without needing to be told how," DeepMind research scientist Oriol Vinyals wrote in a blog post published Friday (4 November). "Games are the perfect environment in which to do this, allowing us to develop and test smarter, more flexible AI algorithms quickly and efficiently, and also providing instant feedback on how we're doing through scores.

"StarCraft is an interesting testing environment for current AI research because it provides a useful bridge to the messiness of the real-world. The skills required for an agent to progress through the environment and play StarCraft well could ultimately transfer to real-world tasks."

Since its release in 2010, StarCraft 2 has become one of the world's most popular eSports that requires tactical thinking, quick reaction time and strategic gameplay.

"Players must send unites to scout unseen areas in order to gain information about their opponent, and then remember that information over a long period of time," Vinyals wrote. "This makes for an even more complex challenge as the environment becomes partially observable - an interesting contrast to perfect information games such as Chess or Go. And this is a real-time strategy game - both players are playing simultaneously, so every decision needs to be computed quickly and efficiently.

"An agent that can play StarCraft will need to demonstrate effective use of memory, an ability to plan over a long time, and the capacity to adapt plans based on new information."

DeepMind will work closely with the makers of StarCraft 2 to create an API featuring a new image-based interface that will allow AI to play directly from pixels and allow bots to take control of individual units. The interface will churn out "simplified" low-resolution images for the map and minimap used by the AI. An option to break down the features into different "layers" such as terrain heightfield, unit type, unit health and others will also be included.

The DeepMind team is also working with Blizzard to design "curriculum" scenarios to provide increasingly complex tasks for researchers to use to "get an agent up and running, and benchmark different algorithms and advances." Researchers will also be able to create their own tasks using StarCraft 2 editing tools as well.

Earlier this year, Google DeepMind's AI software AlphaGo defeated Go world champion Lee Sedol in a historic victory for AI research and development.

However, DeepMind does note that it will be a while before its AI system can take on a professional human player at StarCraft 2.

"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," Vinyals wrote.