> The true mark of AGI is when a company accepts liability and doesn’t bury “for entertainment purposes only” deep in their TOS.
I don't think so. It is easy to imagine the following (currently only fictional) scenario: the AGI does give perfectly correct answers (in a suitable sense), but some people in power consider these answers to be too dangerous, so they sue the company behind the AGI on terms of liability (i.e. the company is liable if the AGI gives answers that those in power don't like and which these people consider to be too dangerous for the public to know).
This doesn't disagree with the poster above: they're saying that taking liability is a sign of belief in AGI. You're saying that lack of liability doesn't mean there's no AGI. Logically these two are not exclusive. p => q doesn't mean q => p.
If the system cannot adjust its answers to the role it's currently serving, then it would evidently be significantly less intelligent than a human.
> If the system cannot adjust its answers to the role it's currently serving, then it would evidently be significantly less intelligent than a human.
Or the system is very smart and recognizes the insanely deep "logical gaps" in the rules of the role that it is to serve.