It is hard to make robots (including telerobots) that are useful, and harder to make autonomous robots that are general and robust. Current robots are created using mathematical models, planning frameworks, reinforcement learning, and manual programming. These methods do not lead to the leaps in performance and generality seen with deep learning, generative AI, and foundation models (FMs). Today’s robots do not yet learn to provide home care, to be nursing assistants, or to interact with people and do household chores nearly as well as people do.
Addressing the aspirational goals of creating service robots requires improving how they are created. Mainstream FMs are not created by agents learning from experience, doing tasks in real world contexts, and interacting with people. They mostly do not learn by sensing, acting, doing experiments, and collaborating. Future robots will need to draw on such experiences in order to be ready for robust deployment in human service applications.
This essay focuses on what human-compatible service robots need to know. It recommends developing experiential (robotic) FMs for bootstrapping them.
Publications
Stefik, M. (2024) What AIs are not Learning (and Why) (16 pages) arXiv https://arxiv.org/abs/2404.04267