cousin_it comments on Complete Class: Consequentialist Foundations

[cousin_it]

I have a recent result about value learning in UDT, it turns out to work very nicely and doesn’t suffer from the problem you describe.