Look, sometimes you’ve just got to do things because they’re awesome.
But would you feel comfortable with that maxim encoded in an AI’s utility function?
For a sufficiently rigorous definition of “awesome”, why not?
If its a terminal value then CEV should converge to it.
Look, sometimes you’ve just got to do things because they’re awesome.
But would you feel comfortable with that maxim encoded in an AI’s utility function?
For a sufficiently rigorous definition of “awesome”, why not?
If its a terminal value then CEV should converge to it.