> This clearly implies that they believe ChatGPT models are AGI and are now willing to say it out loud.

Well, the models are smart enough to point out why this is wrong.