Inner Monologue: Embodied Reasoning Through Planning with Language Models (innermonologue.github.io)

This is ridiculously cool. From the citations in the full paper, it seems like they're using PaLM as their LLM; makes sense, since it's Google research. But man, we're now in the age of, basically, robots that can think about their surroundings, reason about what needs to be done, figure out how, and then execute the plans, including replanning when the environment or request changes, and communicate every thought along the way back to humans in our own natural languages.

I'm nearly 32 years old. Please, please, if there's any divine guidance determining fate, please let us have adaptable and generalizable personal humanoid robots before I die.