>Arent LLMs robust enough to learn on their own via interacting with the world? As far as I know current LLMs are entirely static once trained, they don't lear

>Arent LLMs robust enough to learn on their own via interacting with the world? As far as I know current LLMs are entirely static once trained, they don't learn at all in runtime.