As long as LLMs are stateless systems I'll be having a hard time ascribing any mindfulness to them. For "agents" this state is still in the environment, so no difference there.
Also interesting: How easy it is to ascribe human qualities that we long refused to grant animals at all to a machine.
