AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Importance of Sandboxing Expected Value
Expected value can have this tendency to produce like really large numbers in some circumstances where you're like, well, it's so big it should trump all the other models because it just sort of dominates them. I think that can kind of bound expected value in a way that it's like, okay, no matter how good this expected value is, if it's looking bad on every other front, I probably shouldn't do it. But does Bayesian thinking get you better results? That's actually a different question,. Well, I'm a huge fan of instrumental rationality as opposed to kind of epistemic rationality and that being right about the world is great, but only so far is that let