>>11187665the technical terminology is "Strong AI" : AI that can endlessly and completely reprogram itself according to new rules it fixed autonomously.
i was at a conference of an old french researcher who has been working in that field for 50 years, long story short, progress is not great, but so is interest. we basically haven't got a fucking clue on how to do this, we have neither the theoretical framework, nor the intuition or examples on which to extrapolate. the guy said he'd be happy if there was any kind of advance in the forthcoming 100 years.
so yeah, we are basically way too far to make any kind of informed guess, in all likelihood, this won't happen before at least a few centuries.
