Nothing self-aware about it, it's just LLM output. I'm not trying to be flippant, I just feel really strongly that you shouldn't use words like "self-aware" about an LLM.
It feels like screaming into the void sometimes. People don’t know / don’t want to know. Learned people, smart people, makes no difference.
with more tokens it probably could be. you dont need the real tool call response, if the likely tokens after the tool call(s) is that the user is unhappy, it could predict all those tokens, then have a secomd model predict whether the tool call should be made or not