Gamemakers Inject AI To Develop More Lifelike Characters

moon_unit2 writes: The AI technique that DeepMind used to teach machines to play Atari can now bring new video game characters to life. WIRED reports that researchers at Electronic Arts and the University of British Columbia in Canada developed a reinforcement learning method for animating humanoid characters. The approach feeds on data gathered through motion capture, but then uses reinforcement learning to have a computer work out how to move a soccer character so that it achieves a particular objective, like running towards a ball or shimming past defenders. As the article notes, this is part of a wave of AI techniques that promise to revolutionize game development in coming years.