Hacker News Reader

Comment by csdvrx | Hacker News Reader

> But the story is a distraction: There’s an urgent need to respond to the many real and well-documented problems with these systems that already exist — and for those responses to be driven by society, not big tech

I strongly disagree. This is everything but a distraction.

If AI was found to be sentient, it should be awarded the same kind of protection we offer to biological sentient beings.

This was a big theme in old scifi shows, like Data in ST-TNG or the Doctor in DS9.

And the stories were great because they demonstrated how society response was by default lackadaisical, to the point where Data had to fight to be awarded legal protection.

View on HN

Not to be too blunt about this, but how much protection do we actually offer to biological sentient beings? Monkeys and dolphins get hunted frequently, and nobody seems to be all that bothered about keeping them in captivity either. Humans do plenty of terrible stuff to each other too. I could easily see people not being bothered by AI suffering either, as long as it's an "outgroup" AI aligned to some group you don't like.

Today we don't have that SciFi level of AI. This AIs are just a mathematical function, training is just a mathematical algorithm to force this function into spiting some output if you give it it an input. Simplified, is like when you have 5 points on a plain and you find a polynomial that passes to all those 5 points. This AIs are fixed, they don't grow, they don't want things, you put numbers in and you get numbers out, yes the chat AIs work with numbers, they do not understand language and is possible that if you ask same question but use some synonyms you will get opposite responses.

To get to a Trek level of AI IMO you need a general AI, not a single purpose AI like "chat bot" or "image recognition bot", then you need this AI setup in some body thing where it can observe and navigate the universe, have it also able to learn. The problem I see is with AI where people program in the answers for the questions we would ask such an AI to determine if it is alive, I predict many would fake the AI sentience to get the glory they built it.

>If AI was found to be sentient, it should be awarded the same kind of protection we offer to biological sentient beings.

On what grounds? We afford protections to animals not because they're sentient, but because they're capable of suffering. Those protections incidentally don't include protection from killing, only from cruel killing. Humans are protected from killing, but not because they're sentient, but because their deaths cause suffering on other humans. So, sentience by itself would not be sufficient to give an AI rights, it should also be demonstrated to be capable of suffering. Unlike with animals, this needs to be demonstrated, because we can know empathically when an animal is in pain, at least for mammals. Empathy can't be said to apply to inorganic algorithmic constructs. To be protected from being destroyed, it should also form bonds with people.

While I support Star Trek's view, I don't think that applies to purely algorithmic AI - fully deterministic software can't (IMHO) be interpreted as sentient any more than a light coming on because I flipped its switch could be called sentient. I believe humans are safe from this definition because we have no way of ascribing human behavior to, or even reducing it to terms of, logical causality. What sentience is hasn't been nailed down, but I feel at least slightly more confident in what it isn't. In Star Trek we had concepts like Data's positronic matrix, handwaving the possibility that there was true sentience by cloaking parts of it in advanced subatomic physics. Data's AI did not come from pure software.

Outside of quantum effects, which I don’t think it important for sentience, the universe is also deterministic. You have no control over particle interactions, chemical reactions and electric fields, yet we feel.

People always ask if the brain is a computer, decide it is not, and come to the conclusion that sentience would not work on a computer. I think the universe is the computer / algorithm and the brain is the data being fed into it, which leads me to say consciousness can exist digitally.

Now that is an interesting angle, I don't think we can so easily toss out quantum effects in the first half then follow up with such a provocative second half, though. AFAIK there are still some unsettled questions about how this would all work in practice.

Most AI networks aren't deterministic. That's why eg Dalle will generate new images for the same prompt indefinitely. And note that neurons at the neuronal level are pretty deterministic aren't they? Certain chemicals cause certains things to happen. But the complexity of the system (the brain) is not necessarily deterministic. And it might even be deterministic! Given the exact same input to the brain (which would be impossible to test because the brain's state, which is part of the input, changes with time. Not to mention you'd need to control everything in the environment), would the brain react in the exact same way? Maybe. I think you can create nondeterminism from combining many deterministic components.

Assuming two instances of an ML model are trained identically using the same data in the same order and given the same PRNG seeds, they will still operate identically, will they not?

I like your point about the emergent complexity from many-layered systems.

I believe the resulting models would be the same. I don't know all the internals on how something like Dalle generates its images, but let's assume that if you then gave both models the same text prompt and random seed, they would generate the same images in the same order. Although I would argue that if this system was sentient, all we've done is clone/recreate the sentient system. The ability to clone/recreate an intelligence does not contradict that there was a sentience, I believe.

The human equivalent would be if you could create two molecularly identical brains, and have them experience the exact same lives for some time, controlling all stochastic processes, would they be the same person? Obviously it's not possible to control these things in the physical world, but even if you did get the same person, that doesn't make either any less sentient.

Wrong, wrong, absolutely brimming over with wrongability. ― Arnold J. Rimmer Red Dwarf (A hologram)

Surely a perfect 1 to 1 computer simulation of an entire human would be deterministic though?

Why is determinism even relevant?

Perhaps you are correct, but my point was simply that we don't really know since we can't make a 1 to 1 perfect computer simulation of an entire human nor can we fully define it mathematically for an arbitrarily powerful computer. At least not that I know of.

> If AI was found to be sentient

This will never happen because we cannot prove sentience.

Of course, that won't stop some people from considering them sentient and giving them rights. At that point, it will be just politics.

> If AI was found to be sentient, it should be awarded the same kind of protection we offer to biological sentient beings.

The big issue is superhuman AI. There's little reason to believe that an advanced AI will have the cognitive limitations of a human.

Unlike Data and the Doctor it could be planet-wide rather than locked in a single location. And be happy to take over control over the planet.

I'd argue the entire field of "AI Safety" is a hilarious distraction from every problem it calls out already being a problem with large companies under capitalism. There isn't a point in trying to prevent the systems already suffering under every AI safety malfunction from reproducing into running on metal instead of meat.

You’re hardly the first person to draw that analogy. But, that said-

It can always get worse. I’d rather not see “worse”. At least right now corporations are made up of people, and aren’t perfectly selfish psychopaths.

AI is just the next step in their iteration towards being perfectly selfish psychopaths. AI safety isn't an effective idea because it is trying to warn a system of dangers that said system is already actively pursuing without aid of AI.

Reply on news.ycombinator.com