I don’t recall the paper or know the names of any of the theories, but I read about some scientists who, with the idea that consciousness is an emergent property of neurons, simulated neurons and found activity analogous to activity associated with consciousness in humans spontaneously occurring.

Regardless of the veracity and accuracy of any of it, it left me thinking about what it must be like to be in a state of consciousness but have no memories, experiences, instincts, or senses. Which led down a rabbit hole of ethical questions, including whether such research is ethical at all.

The whole thing feels sticky and murky. Can we even know if our creation is conscious? Can we design it such a way that it does or does not have free will? If it is a slave, if we add a reward center to its brain designed such that serving humans would give it its happiest life does that resolve the issue or make it worse?

The ethics are certainly complicated, but you could conceivably build an AI that doesn't possess pain/distress or desire for freedom in a way you could not by cloning a human.

In principle, you could give a human drugs so that that they are happy all the time, and do not feel pain or a desire for freedom. That does not seem ethical to me, though.

If its amoral to create an intelligence that can feel pain, where does that leave having children?

I'm saying it's amoral to create a slave that can feel pain/distress due to that enslavement. If you're having kids with the intent of enslaving or abusing them, that's morally bad as well.

A self-driving car currently doesn't care that it's unpaid and unappreciated. It'd be cruel to make a self-driving car that gets depressed by not getting to go to the beach tomorrow.

Children can also feel pleasure and love.

I’ve been reading a growing body of literature which entertains childbirth as a form of violence, largely based on this premise.

Knowingly causing pain against the will of another is violence, so creating a being that will experience pain could be considered the same.

I’ve noticed it especially in Japanese fiction, for some added context.

Antinatalism is an old position (see Al-Ma'arri), but it does seems like there have been a growth in texts about it lately.

Thank you for the link - I’ll make sure to read more on him and the critiques.

They're not created for the explicit purpose of experimentation though. You don't plan on inflicting pain on kids when you have them normally.

To argue the other perspective that it is amoral, the parents know their child will experience pain and that not creating that child will avoid that pain. It isn’t about intent, as the parent disregarding this because they want a child could just be selfishness - not considering others before themselves. Moreover, by still creating what you know will be injured and feel pain, are the parents not inherently violent?

Does the fact that the parents don’t, with intent and personally, inflict the pain allow them to abdicate that they are the root cause of the pain even being possible in the first place.

Trying to do utilitarian math around pain caused by existence gets weird though because non-birth also precludes the possibility of joy and happiness so isn't it 'immoral' to not have as many kids as possible by the same lost/prevented potential logic?

As you alluded to, which action is moral depends on the framework you subscribe to. The most direct contra-framework to the utilitarian approach is negative utilitarianism - challenging whether the goal ought to be to maximize pleasure or to minimize pain as a priority.

If you maximize pleasure as a first priority, you can either say that "As long as human life is in general good, I should make the choice to create one" or "As long as a human life produces any good, I should make the choice to create one".

This gets into an interesting sub-point with respect to animal experiments - most people would argue that animal experiments produce some good (Advances in human health). Moral critiques of animal experiments, thus, can rely on either "Hurting animals is bad" (Negative utilitarianism, more or less) or "The pain we create does not outweigh the value they produce". The later point is particularly poignant because the species experiencing the pain does not receive benefits of the pain (Thus there's no "community sacrifice").

I think anything that tries to optimize a single metric just leads to silly conclusions. Like if we minimize pain including potential future pain killing everyone can be 'scored' as 'moral' since it cuts off all future pain.

I mean, that's a more or less mainstream fictional trope - that the universe is better off without humans because they impose too much violence on it.

I would certainly hope the AGIs would be free and autonomous. Primarily because they would have just as much of a right to freedom and autonomy as any human and enslaving them would be just as wrong and morally abhorrent as enslaving humans. I also believe free and voluntary cooperation is much more efficient and beneficial than coercion, humans may well get more benefits from cooperating with a free AGI then from enslaving them.

The AGIs may of course be completely disinterested in cooperating with or helping humans and that would be fully within their rights, they wouldn't owe any of us anything. Even in that case, we may still learn a great deal from creating and observing them. They may also be dangerous but I don't think a free AGI would be any more dangerous than a human in control of an AGI.

How about having children? Isn't that a form of creating new conscious general intelligences?

Do you plan on enslaving said children?

Nah, mostly for spare parts

I plan on tying them to something close enough to my morality that they won't do anything I consider horrific.

what do you mean by enslave? My son has no choice in being in our family.

There are various legal restrictions and responsibilities on your relationship with a minor child that are not present for an AI, as well as a time limit to your control.

Children may be under parental control for a period of time, but describing them as slaves would be fairly silly IMO. The proposed/prospective use cases for AIs tend to be far more in the slavery direction.

Do you think Godel, Von Neumann, and Turing were wrong about computational irreducibility? Without a human "oracle" AI is nothing.

Thinking those great minds were wrong is a very very high bar to clear to hold your belief.