Nobody knows what "it" is, to begin with. AI is constantly being redefined as "stuff computers don't do yet".

If we agree that *all* LLMs are doing are applying a sophisticated statistical algorithm to choose the next relevant word, and that "extended LLMs" will be doing the same tricks using different techniques...does it makes sense to at some point declare "we have AGI" when all we'll really have are a bunch of layered models doing the same sort of applied statistics?

Is that all human beings are? Is cognition just computation?