factual

Objectives Need Not Be Explicitly Represented

AI systems like AlphaGo do not store their objective explicitly; the win-objective is built in only during training as a reward signal, while the deployed system uses an approximate learned predictor and incomplete search as a proxy, and in general explicit representation of an objective is only needed when that objective is changeable, just as a fixed-destination navigation system needs no explicit destination.

factualpending

Speaker

Stuart Russell

Evidence Quote

“alphago is something that was trained with the objective of winning the game which learned an approximate predictor for whether a given state is likely to lead to a win and then does an approximate search”

Source

153 - Possible Minds— Making Sense with Sam Harris
Created: 6/14/2026, 2:27:33 AM

Your notes

Loading notes...