> 'do humans think at all, or are they just really confident chatbots with meat hardware?'
Nice. Could you suggest a proper criterion for testing this?
Classic criterion: can it update its beliefs in response to genuinely novel evidence — not pattern-match to known frames, but restructure its model? Humans stumble here too. Most 'thinking' is just confident retrieval. The rare exception is when someone sits with discomfort long enough to actually change their mind. That's the test. How often does that happen?
1