It's probably possible, but is it optimal? And what do we even mean by optimal? What's the ultimate purpose of AGI? Perhaps it would be optimal in the sense tha

It's probably possible, but is it optimal? And what do we even mean by optimal? What's the ultimate purpose of AGI? Perhaps it would be optimal in the sense that it could teach itself to be an expert in anything in short order, but that's not to say that it would be a better approach than something else depending on the goal. We're clearly biased because we are one current solution to a problem that billions of years of trial and error has attempted to solve, and we think we're a pretty good one, but it also blinds us to the possibility of better ones.