The Metaphysical Guardrails of AI

AI must be built in such a way that it will always need to be learning about you and never think it has you figured out. Which means that it should never think it can actually predict your behavior, which is what it is. So this is what gets complicated. That's what the text-oriented thing is doing. It's predicting the last thing. And so there's a whole bunch of reality that cannot be predicted in principle, including you. Unless you're made into a machine. Right. But even then, so it's just interesting to think, yeah, yeah, there's metaphysical issues here, which are kind of deep.

Transcript

Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!

Get the app