Last Week in AI

Every week, my team at Invector Labs publishes a newsletter that covers the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:

From the Editor: Mastering Go, Chess, Shogi and Atari Without Knowing the Rules

Gaming is one of the areas in which AI has shown the most progress in the last few years. From checker to Go to StarCraft, AI programs has regularly achieved superhuman performance and shown signs of creativity. However, in all scenarios, those AI programs started by being trained in the rules of the game. What would happened if that wasn’t the case? After all, there are plenty of scenarios in which we are faced with a brand new environment that we need to learn as we go along. Humans are extremely good at this but can AI agents do it?

This week, DeepMind published a research paper unveiling MuZero, a reinforcement learning agent that mastered games like Atari, chess, shogi and Go without knowing the rules upfront. Using a technique known as model-based reinforcement learning, MuZero will first learn the dynamics of the environment( rules of the game) and then learning the best strategies conducive to a successful outcome. The principles of MuZero applied to many scenarios in robotics or…