You could run Parfit’s hitchhiker pretty easily (you literally have access to the function that computes the AI’s behaviour, and so can estimate the chance it will pay the driver). So it really looks like ES will learn to pay. EDT does not pay the driver after it learns that it has already escaped the desert; paying would no longer be evidence of survival.
I think that only gets you EDT.
You could run Parfit’s hitchhiker pretty easily (you literally have access to the function that computes the AI’s behaviour, and so can estimate the chance it will pay the driver). So it really looks like ES will learn to pay. EDT does not pay the driver after it learns that it has already escaped the desert; paying would no longer be evidence of survival.
Yes, thanks. In retrospect I was considering the slightly cleverer strategy “do what’s worked in the past”, which fails on Parfit’s hitchhiker.