Yeah, the food that is served in fast food restaurants, and arguably a lot of society, basically wireheads our reward centers, and to a large extent is why obesity is such a huge problem in the modern era.
Obesity is the first example of real life wireheading, at least in a weak sense. So now that I think about it, I think TurnTrout is too optimistic about RL models not optimizing reward.
Yeah, the food that is served in fast food restaurants, and arguably a lot of society, basically wireheads our reward centers, and to a large extent is why obesity is such a huge problem in the modern era.
Obesity is the first example of real life wireheading, at least in a weak sense. So now that I think about it, I think TurnTrout is too optimistic about RL models not optimizing reward.