Read on Twitter
I think only Wittgenstein could set right the utilitarians that start to get so concerned about pleasure and pain in abstract that they start to worry about reinforcement learning (this is small subset of VERY through STEMy utilitarians, but it's a real issue raised, believe me)
Wittgenstein because I think he has relevant insights about meaning in context that would apply here. What do we *mean* by pleasure and pain in the sense that it matters to us? I think we have in mind the context it plays in our form of life, in our meaningful life activities
The concern about a reinforcement learning algorithm represents the reductio ad absurdum of this kind of utilitarian 'out of context' treatment of ethic. Of course, some are willing to bite the bullet. But I think most would not, and most utilitarians might not even go that far.
I leave the possibility open that reinforcement learning algorithms could have some kind of qualia. That is another question entirely. But there's no reason to think it would be ethically relevant qualia.
And one can still be concerned about animal pleasure, pain IMO, I am to some extent. But why are we in the first place? Out of an abstract concern for pleasure, pain? Or is it because we relate our form of life to theirs, being similar enough as biological organisms.
This is why we tend to be less and less sympathetic the more different it is IMO, from insects and fish all the way to plants. I don't know much about whatever qualia is in that range. But the sort of ethically meaningful experiences, life we are concerned barely extends to them
This is an anthropocentric view of ethics! And that's a good thing, and the only type of ethics really possible from an anthropomorphic standpoint. The 'human' it is worth putting is something greater than just a biological form as it has culturally and ethically understood tho
