Google’s DeepMind AI learns to play StarCraft

    145

    No, playing video games isn’t a waste of time. At least not for Google’s artificial intelligence machine DeepMind. Google just announced they’re partnering with Blizzard to create an AI-friendly version of the popular game StarCraft so that DeepMind can play the game. It’s not just so the AI machine can hone in its zerg skills, Google hopes this will help it learn how to solve problems that come with real-world “messiness”.

    “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,” the DeepMind team wrote in a blog post. “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.”

    Some might recall Google’s DeepMind technology from its historic Go match against the world’s top Go champion Lee Sedol. In March, the AlphaGo won 4 out of 5 games against the human player after programmers gave it the principles of the game and then let it self-improve through practice.


    But the game of StarCraft brings some new challenges missing in turn-based games like Chess and Go. For instance, parts of the map in StarCraft are obscured to the player and information is incomplete. This means players must make the best decision they can under pressure against competitors who aren’t waiting for their own turn to make counter-moves. So the AI will be challenged to make efficient decisions as quick as possible in a setting more like the everyday world. Researchers are hoping to use the game to test out different theories on how machines can learn to tackle complicated systems and environments.

    For now, it’s all fun and games. But the skills DeepMind could develop by playing an elaborate game like StarCraft has real-world applications. For instance, it will need this level of strategic and adaptable thinking to perform such tasks as managing air traffic at busy airports or working out elaborate shipping and manufacturing networks. Or it could make the even smaller move over from thinking out fantasy warfare strategies into real-life ones.


    Avatar
    Kelly Paik writes about science and technology for Fanvive. When she's not catching up on the latest innovations, she uses her free-time painting and roaming to places with languages she can't speak. Because she rather enjoys fumbling through cities and picking things on the menu through a process of eeny meeny miny moe.