New MIT Research Suggests That Training An AI Model With Mathematically “Diverse” Teammates Can Improve Its Ability To Collaborate With Other AI It Has Never Worked With Before

The prevalence of superhuman artificial intelligence (AI) in competitive games such as chess, Atari, StarCraft II, DotA, and poker is growing. Recent advances in deep reinforcement learning contribute significantly to the success of competitive AI. In contrast, cooperative AI, in which autonomous agents must interact with humans or separately-trained agents to achieve a common goal, […]

The post New MIT Research Suggests That Training An AI Model With Mathematically “Diverse” Teammates Can Improve Its Ability To Collaborate With Other AI …

ai ai model ai paper summary ai shorts applications artificial intelligence carnegie mellon university country editors pick mathematics mit reinforcement learning research staff tech news technology training university research usa