LLMs are fairly capable physical agents already. Nothing large about the assumption at all. Not that a robotic threat is even necessary. https://tidybot.cs.pri

LLMs are fairly capable physical agents already. Nothing large about the assumption at all. Not that a robotic threat is even necessary. https://tidybot.cs.princeton.edu/ https://tidybot.cs.princeton.edu/ https://innermonologue.github.io/ https://innermonologue.github.io/ https://palm-e.github.io/ https://palm-e.github.io/ https://www.microsoft.com/en-us/research/group/autonomous-systems-group-robotics/articles/chatgpt-for-robotics/ https://www.microsoft.com/en-us/research/group/autonomous-sy...