While I agree with the hidden statement of the utility in online reinforcement learning here, it should be pointed out that for some snapshot of a system, which

While I agree with the hidden statement of the utility in online reinforcement learning here, it should be pointed out that for some snapshot of a system, which may have already been trained with a large amount of data - the structure of the learned space and the inferences within it seem like a reasonable definition for 'understanding'. I don't know many people that would suggest that knowing about ones family structure, such as their mom/dad/uncle, and how their history relates to them, is required to be _completely reconstructed_ every time they interact with their environment somehow from first principles. Online reinforcement learning can have large merits without resorting to stating it's required for learning. Just as self awareness can occur independently of consciousness can occur independently of intelligence is independent of empathy and so on. They are all different, and having one component doesn't mean anything about the rest.