DeepMind's new agent can learn complex games. That's a big step toward AI that can navigate the real worldAI agent SIMA can learn to perform diverse tasks in virtual environments using natural language directions.DeepMind aims to build AI models for real-world tasks based on SIMA's behavior in unfamiliar environments.
Google's new AI will play video games with you - but not to winSIMA AI agent trained to play games like a humanNot meant to replace game AI, but as an additional player in the party