The way some (!) artificial intelligence works is by reward functions. Essentially, the neural network gets a problem to solve, responds with an action and then we give a “score” back to it. Say, bowling. Each pin gives +10, missing the throw is -50, and by letting it repeat and repeat it will try out all kinds of angles and speeds and spins until it settles into solution that gives it maximum points.

I believe with humans it's the same. Humans constantly train their neural networks (learn) and pain and suffering are a conscious expression of reward mechanisms being triggered in your brain. All pain is aimed at discouraging you from attempting that strategy again, and all pleasure is meant to encourage you to do the same thing again.

Inside the human brain, you could, in theory, boil everything down into the result of a reward function, and that reward function is comprised of the many little things that human brains consider to be conducive to their survival and the things they consider to be deducive to their survival. Snakes and getting bit are deducive and therefore humans, even babies, are naturally averse to snakes and feel pain (ie. the reduction of a number in the reward function) where it takes conscious effort and training to overcome that fear (though it's easier to do with babies who have only what little fear of snakes can be coded into dna as opposed to real life negative experiences).

In that vein, we need to talk about local optima. Back to bowling, bowling isn't as easy as selecting angle, speed and spin and then hitting a button, it's a complex set of over 600 muscles to coordinate all at once, and an artificial intelligence doesn't know what an arm is or how physics work, it just sees a list of 600 “motors” it can contract and relax and then after a while gets a number: “You did X good”. So, like with humans when trying to walk, you can't really tell it “just hit the pins” because then it will never be able to figure it out, you have to design your reward function in such a way as to reward behaviors that are likely to lead to the desired result, not the desired result directly. So, for example with bowling we could integrate a factor into our reward function that rewards *speed*, because in bowling you want to throw the ball hard. The problem with this is that depending on how your reward function is set up, it might actually be more lucrative to aim for just raw speed than to bother with any of the other factors in the reward function. So, your AI might end up throwing the ball hard and fast… in the opposite direction of the alleys because there is a long hallway where the ball can roll without collision for a long, long time without interruption. This is how drug addictions work. Living a life without drugs is generally healthier but getting an immediate reward is much easier and faster to do, and the brain doesn't even know how many “reward points” it is missing out on if it did life “correctly”.

The problem with this, and why it's called a local optimum, is because trying to nudge the AI to a better solution will always require the AI to try things it will, initially not be as good as as the thing it's already doing. For example, if you tried to very strongly “encourage” the AI to throw into a particular direction then with its skillset and strategies it will not be able to score as many points as it would be able to do with its current strategies and skillset. This is why “delayed gratification” is such a big buzzword, doing your homework or quitting drugs is extremely painful from a reward function perspective and since we naturally gravitate towards maximizing our reward function the easiest thing to do is to just not do homework or to keep doing your drugs.

This is the same reason why hitting your child or talking down to is does NOT work. It doesn't do anything. If a kid just keeps doing a thing you would rather it did not then it's because it found a local optimum with the strategies it has available to maximize its reward function ('gratification'). Unless you make that particular spot of the reward curve a massive pluging black hole that is so immediately punishing that even the territories around that behavior are less painful that what it's doing right now then you're not helping (and if you do that, you're just a traumatizing monster). What you need to do is to find where on the reward curve the kid is, which strategies it uses to get there, then shape the world around you in a way such that an immediately neighboring strategy is more conducive to its reward function than the current one, because then it will automatically gravitate towards that. That way you can slowly guide it into new strategies. Basically, to reach an even higher peak you first have to descent from your current mountain, and you cannot slap a kid until it falls off the mountain.

https://www.youtube.com/watch?v=EWjUY_3ubf4

“AI Invents New Bowling Techniques” by b2studios if you want to get a concrete example of reward functions and local optima.

Oh, right. So, whenever you do something that is conducive or deducive to your survival, the human activates the reward mechanism, which propagates throughout the brain and triggers neighboring nerve cells, making you perceive it consciously.

Anyway, the point is that this is ultimately related to utilitarianism. The goal of utilitarianism is to maximize our returns from the reward function, for all humans. A future machine might be able to inject gratification/utility/reward straight into a human brain, but right now we just have the limited world of our physical existence to work with and that's how we get back to utilitarianism. The ethical thing to do is the thing that maximizes the reward function as much as possible, for as many people as possible. Denying people the right to have gay sex reduces the reward function for millions without proper justification. Forcing people to take the vaccine increases it, though perhaps not for yourself. Simple stuff (you'd think).

So, from a utilitarian perspective it is in your interest to give people the right to have full autonomy over ther bodies, including the autonomy to make it un-dynamic (dead). At the same time, from a utilitarian perspective it is in societies interest to stop people from committing suicide because, well, there always is the possibility for them to find back to a life they would genuinely enjoy, no matter how much pain that involves. And if you look out into the world, that's how most governments are structured: You are in your complete right to kill yourself if you so want, but society as a whole is legally required to stop you from doing it. In the same way that you can get charged for driving past the scene of an accident, not stopping someone from committing suicide is seen as a societal responsibility for purely utilitarian reasons. The difference between an accident and suicide is that suicide has at least some degree of consent in there and only directly (!) affects the victim, whereas an accident can directly (!) affect otherwise not directly involved people. In other words, in a suicide the damage is contained (disregarding the suicide victim's family), whereas in an accident more than the person at fault can/will be harmed. This difference is the reason why there are subtle differences in country legislations, where for example in the Netherlands assisted suicide through doctors is actually a thing while completely illegal in Germany.

So, what IS it about the families of suicide victims? Aren't the suicidees reducing their reward functions?

Well, yes, but that's not your fault. As the person committing suicide, you're not the reason they are feeling pain, they are feeling pain because they chose to get attached to you/give birth to you in spite of the fact that things like suicide *can* happen. They knew the risks and accepted it, so they are accepting your potential suicide as well. A world where you are completely beholden to the emotions of other people would be terribly immoral for one, but also completely dysfunctional because you have to consider everyone you could ever affect with your actions - and everyone has a different opinion on everything, so you would never be able to *do* anything. You would not be allowed to kill yourself (according to some), but you would also never be allowed to be alive (according to others). Don't quote me on this, but getting attached to someone is the fault of the person doing it, not the person they are getting attached to.

Unless, of course, you explicitly tell someone that you won't but then do it anyway, because by the time you said you won't there is a commitment on your side and people rely on that. Until then, however, nobody can prescribe life onto you, they can only justify physically coercing you into it.

Nihilism isn't the answer to anything. Nihilism is the philosophical bedrock of reality. You think, therefore you are, and for some reason you care about your existence but there is no foolproof way of actually justifying your value to anyone. Things exist and that is it, period. From this perspective, happiness and pain also do not mean anything but you will quickly find that if you ask around, most people would *not* want to live in a world where happiness and pain are meaningless because it would lead to a lot of pain. Which is, I know, kind of pointless, but it doesn't matter. Yes, beneath the utilitarian pragmatic shell there is a foundational layer in absurdism in me that believes that happiness and pain are absurd ideas without any actual merit – because merit itself isn't a thing. Atoms exist, maybe, and they can change places, over time, maybe, but everything else are dumb concepts in brains. You are right about that.

But this thinking doesn't work on a human level. Your emotions, however absurd they are, are real. Your pain and your pleasures are real. Your perception of reality (though not necessarily the reality you perceive) is real. They are all you ever had and will have, so to you as an observer they are the only thing that, well, “matter”. If you want to improve lives for humans on earth rather than “conscious things as a philosophical concept” then you have to take into account human qualia, and this - at least for me - has lead me down the road of utilitarianism.