Way beyond AlphaZero: Berkeley and Google work shows robotics may be the deepest machine learning of all

The vast majority of artificial intelligence has been developed in an idealized environment: a computer simulation that dodges the bumps of the real world. Be it DeepMind’s MuZero program for Go and chess and Atari or OpenAI’s GPT-3 for language generation, the most sophisticated deep learning programs have all benefitted from a pruned set of constraints by which software is improved.

Read More