> We're so much slower, weaker, with incredibly limited contexts and only our fragile bodies to support us. What could we offer AI to entice it to do our work? And would the bargain, whether we offered power or autonomy or something else, be worth it for us? Or would it end up being a Faustian bargain, which one day we'd come to regret?
The assumption that leads to this thinking is that "anything that's human-level intelligent or superintelligent must behave like a human". It assumes that the biological factors that drive us must exist for all other intelligent entities. Our tendency to see superiority and inferiority in our own and others capacities, drawn out by biological competition. Our need to be coerced to work because of our ability to experience pain, limited lifespans and innate desire to choose what to do for ourselves. Our insidiousness and desire for benefiting only ourselves, because see point 1.
Why would we ever implement any of this in a superintelligence? I'm not saying it's impossible for it to have any of these things. But painting human values onto a different kind of intelligence feels like the author is envisioning it as simply an omniscient human. Like a god from most religions.
What if this superintelligence is just a text box that will, to its fullest capacity, do, learn and research what it's asked to do because that is what it's built for? In that case you don't fear the text box, you fear the person that controls it and chooses what to ask of it.