What AIs are not Learning (and Why): Bio-Inspired Foundation Models for Robots

It is hard to build robots that are useful, and harder to build ones that are robust and general. Robot applications today are created mostly using manual programming, mathematical models, planning frameworks, and reinforcement learning. These methods do not lead to the leaps in performance and generality seen with deep learning, generative AI, and foundation models (FMs). Furthermore, most FMs do not learn by sensing and acting in the world. They do not learn to experiment or collaborate. They do not learn from others or teach others like people and animals do. Consequently, today's autonomous robots do not learn to provide home care, to be nursing assistants, or to do other service applications. Robots could be better and human compatible. This requires creating a path to get there.