cynar comments on A Microsoft researcher built a goat-powered LLM in Age of Empires II to prove it's not sentient

[–] cynar@lemmy.world 5 points 2 months ago (11 children)

I'm not expecting anything any time soon either. Though I can see someone like musk pumping far too much money into it at some point.

My point was however that the difference is just one of scale. We don't need to predict the firings, just run it and compare it to nature. From what I've read, it behaves like a fly, including walking and grooming itself. This means there is no magic mystical difference between a real fly's brain and a virtualized one.

Projecting further, there is no difference, other than scale, between our brain and the fly. Implying there is nothing mystical about consciousness.

If a human brain can be conscious, then a virtualized human brain can be conscious. If a virtualized brain can be conscious, then so can the computer it runs on.

The question then becomes do we WANT consciousness in an AI, what would it look like, and how can we detect/measure it?

permalink

fedilink

source

parent

hideshow 11 child comments

[–] marxismtomorrow@lemmy.today 3 points 2 months ago (2 children)

Except if this is the case, free will does not exist. If the world is purely deterministic, you've never made a choice. No one has. We are in the middle of a mathematical computation that has been predicted from the start of the universe.

permalink

fedilink

source

parent

hideshow 2 child comments

[–] cynar@lemmy.world 2 points 2 months ago (1 child)

That is pretty much my view on things. I don't like it, but it's what the evidence suggests. However, my internal thoughts still assume I have free will. It's a useful lie.

Discworld's Death put it quite well, in Hogfather.

All right," said Susan. "I'm not stupid. You're saying humans need... fantasies to make life bearable."

REALLY? AS IF IT WAS SOME KIND OF PINK PILL? NO. HUMANS NEED FANTASY TO BE HUMAN. TO BE THE PLACE WHERE THE FALLING ANGEL MEETS THE RISING APE.

"Tooth fairies? Hogfathers? Little—"

YES. AS PRACTICE. YOU HAVE TO START OUT LEARNING TO BELIEVE THE LITTLE LIES.

"So we can believe the big ones?"

YES. JUSTICE. MERCY. DUTY. THAT SORT OF THING.

"They're not the same at all!"

YOU THINK SO? THEN TAKE THE UNIVERSE AND GRIND IT DOWN TO THE FINEST POWDER AND SIEVE IT THROUGH THE FINEST SIEVE AND THEN SHOW ME ONE ATOM OF JUSTICE, ONE MOLECULE OF MERCY. AND YET—Death waved a hand. AND YET YOU ACT AS IF THERE IS SOME IDEAL ORDER IN THE WORLD, AS IF THERE IS SOME...SOME RIGHTNESS IN THE UNIVERSE BY WHICH IT MAY BE JUDGED.

"Yes, but people have got to believe that, or what's the point—"

MY POINT EXACTLY.

permalink

fedilink

source

parent

hideshow 1 child comment

[–] marxismtomorrow@lemmy.today 0 points 2 months ago

Except it's not what the evidence suggests. We have yet to find physical evidence of consciousness. We have yet to find a way to accurately simulate a neuron in any falsifiable way (seriously look into the detractors and counterarguments to the fly study, it's amazing that 3d graphics can trick that many people). But even beyond neurons we cannot accurately link brain activity to behavior in anything but a superficial way. We have no insight into how decisions are made or what ultimately drives them.

More importantly, if the world is deterministic, all meaning in life is gone. Completely. And the second it is objectively proven most people will kill themselves. Any intelligent being would kill itself, since that's the logical next step in the chain of thought and it wouldn't have a choice.

Meaning the universe is flawed if the point of it is to create life, and if there is no point to the universe and it's purely deterministic, we are in a worse hell than has ever been imagined in the totality of human expression. Arguably the worst possible hell there could ever be.

Nonexistence would be the infinitely better option.

permalink

fedilink

source

parent

[–] partofthevoice@lemmy.zip 2 points 2 months ago (6 children)

This means there is no magic mystical difference between a real fly's brain and a virtualized one.

That’s not true. You might simulate fly intelligence, but you aren’t simulating a fly. Flys are an entity driven by their phenomenological experience of the world (as all us nervous systems are). It would be a rather strange thing to say that a fly is only the pattern of behavior you recognize as a fly’s behavior.

Note that flies also have this capacity to self-evolve over generations. There is no single fly that you can point to and say, “that’s the right one. Copy It.” So, your “fly simulations” are always at best a behavioral approximation. Given enough time, say a decade or century, this ought be obvious by the fact that your simulation no longer accurately resembles the most modern flies.

Projecting further, there is no difference, other than scale, between our brain and the fly.

That’s just not true. Consider “What The Frog Eye Tells The Frog Brain.” Put briefly, the eye encodes and transmits semantic information — as opposed to the more common belief that it transmits raw visual information. That said, there a trillions of differences between how that might work in a human versus in a frog, let alone a fly.

I think you’ve discovered “neurons work similarly across species,” which is like saying “thing does same thing when used in other location.” This doesn’t tell you how neurons work to drive that behavior, it doesn’t tell you what it’s like to experience your neurons firing that way, it doesn’t tell you why the neurons were developed that way topologically over time… At best, it helps you develop a timebound understanding of fly automata to neural architecture. That’s without any understanding of phenomenology to neural architecture, and likely without being able to decompose the neural system into any semblance of semantic information processing.

A human is much more complex than a fly. I can’t believe for a second that this approach would work scale to a human’s brain. If you captured and simulated every neuron in the human brain, you’d be left with a feedforward process simulating a particular neural state of a particular person. No feeling, no reflection, no introspection, no semantic meaning. Just a neat toy.

permalink

fedilink

source

parent

hideshow 6 child comments

[–] Jakeroxs@sh.itjust.works 3 points 2 months ago (5 children)

"If you captured and simulated every neuron in the human brain, you’d be left with a feedforward process simulating a particular neural state of a particular person. No feeling, no reflection, no introspection, no semantic meaning. Just a neat toy."

Seems to me you think there's something special about humans, I don't believe there's any proof to that.

Yes its orders of magnitude more complex then a fly brain, that doesn't mean it's impossible to simulate.

There's no proof to a soul, our whole existence is our meat which evolved naturally over extreme timescales via random forces and natural selection, I see no reason to believe we could not do the same with our intelligence in a much shorter period of time comparatively.

permalink

fedilink

source

parent

hideshow 5 child comments

[–] partofthevoice@lemmy.zip 1 point 2 months ago* (4 children)

Seems to me you think there's something special about humans, I don't believe there's any proof to that.

I don’t, though. I get where you’re coming from, because I too often point out other people’s hidden assumptions about things like a soul or magic sauce to consciousness. I don’t subscribe. I follow something closer to a procedural understanding of consciousness.

that doesn't mean it's impossible to simulate.

I think it does. You aren’t going to simulate human consciousness if human consciousness is a procedure. At least, not how you’d simulate stock exchanges.

The difference is simulation versus synthesis. Your simulation would be a feed forward process and humans are not feed forward processes. Humans are phenomenologically driven entities. Your simulation would be a behaviorally consistent model at best — no phenomenology. Sorry to say, but that’s just not the same thing. It wouldn’t work the same way, wouldn’t feel the same, and you definitely wouldn’t have “figured out consciousness” just by building such a simulation. It would be a neat toy, like LLMs, but it wouldn’t be sentient. That’s obvious to me.

There's no proof to a soul, our whole existence is our meat which evolved naturally over extreme timescales via random forces and natural selection, I see no reason to believe we could not do the same with our intelligence in a much shorter period of time comparatively.

Yeah, and I’m really not arguing for a soul. I’m arguing that simulation is subpar to recreation. If you simulated every neuron, you haven’t achieved anything really remarkable — except an awesome benchmark for computer power and a place in the history book. That’s it.

If you want a grand unifying theory of consciousness, you need to understand that phenomenology (the thing science ignores) is pretty damn central to the whole thing. That’s not arguing for a soul. That’s saying, pinch yourself — you are awake and you are not a robot.

permalink

fedilink

source

parent

hideshow 4 child comments

[–] Jakeroxs@sh.itjust.works 2 points 2 months ago* (3 children)

At what point to you would it move from simulation to essentially the same configuration just in a different container?

I still think you're conflating the aspect of life we experience as somehow separate from the procedure our brain takes to present that to you.

permalink

fedilink

source

parent

hideshow 3 child comments

[–] partofthevoice@lemmy.zip 1 point 2 months ago* (2 children)

At what point to you would it move from simulation to essentially the same configuration just in a different container?

Never.

Intelligence is substrate independent, and does not depend on consciousness or phenomenology. Phenomenology is probably substrate dependent, though. Meaning, the meatbag wrapped around our skeletal system really does matter. Phenomenology is also what makes us feel anything at all.

You can make all the intelligence you want, bootstrapping it off our neural architecture. You’re still doing us a disservice if you call it human. It’s more like a new species, with much more intelligence and virtually no phenomenological experience.

But calling it a new species is also arguable, because whether or not it would be “alive” is arguable. I’d just call it a neat toy.

permalink

fedilink

source

parent

hideshow 2 child comments

[–] Jakeroxs@sh.itjust.works 2 points 2 months ago (1 child)

I don't understand why you think it'd be out of the question for it to have phenomoligcal experiences, why couldn't we build a way for it to smell if it has the same pathways to be able to interpret that? How would that be meaningfully different?

Think about hearing aides or other similar "enhancements" wherein were simply adjusting the input in a way and the brain is still able to process it.

I would agree it wouldn't be the same as a literal human, thinking like Fallout 4 style synths, I consider them people (in universe of course lol) even if they're not literally humans.

permalink

fedilink

source

parent

hideshow 1 child comment

[–] partofthevoice@lemmy.zip 1 point 2 months ago*

I don't understand why you think it'd be out of the question for it to have phenomoligcal experiences, why couldn't we build a way for it to smell if it has the same pathways to be able to interpret that? How would that be meaningfully different?

The difference is having a phenomenological experience versus data processing. A phenomenological experience actually feels like something. We have literally no reason to believe phenomenology isn’t deeply dependent on the body. In fact, we probably have more evidence that it does depend on the body. Science doesn’t do a good job at justifying phenomenology though… and you aren’t going to conveniently recreate phenomenology using brute force via a technique that hinges itself on conveniently ignoring subjective information such as how things feel.

If your simulation has the “same pathways” to smell, congratulations—you just remade biological life. You made the nose, brain, and the connections between the two. If you didn’t do that, then you made a fancy copy cat machine. It doesn’t actually smell, not anymore than ChatGPT actually thinks.

Maybe you can give it a device that lets it convert air-born chemical signals into information it can act on. That’s a feature of intelligence though… not really a phenomenologically active experience. If you asked this intelligence machine “can you smell like I do?” I imagine it might respond, “no. I can process chemical information much like you do, but my ability to do so is based on data whereas yours is based on feeling or intuition.”

You literally cant experience the world like a computer would—without personal and phenomenal subjectivity. Equally so, a computer cant experience the world like you would. Even simulated, that’s not the same thing as feeling something.

Honestly, … I’m not saying this machine wouldn’t have applications. It very well may be like a supercharged LLM, for all I know. It’s just not human, no more than a Birch can ever be a Redwood.

I don't understand why you think it'd be out of the question for it to have phenomenological experiences

Addressing that more specifically, I think it’s because I have yet to see a single scientific development in the wake of understanding the cause and nature of phenomenology. It’s ignored, rightfully so. Science doesn’t need to concern itself with phenomenology, at least hasn’t so far.

People have similarly argued that consciousness “arises” when an information machine is sufficiently complex. This claim of weak emergence seems ignorant to me; “arise” is doing a lot of magic handwaving. I’d argue it’s the same for phenomenology — you’re not going to stumble upon how to make a machine feel just by brute forcing more complexity.

Also, we should not be so naïve to think that understanding phenomenology is unnecessary for our goals [only because it’s not in the scientific spotlight]. Phenomenology is central to our experience and every other complex life forms experience (given a CNS).

Slime mold is intelligent, bear in mind. No nervous system, I’m doubtful it “feels” anything, but it’s intelligent. Imagine that you reprogrammed some slime mold to process information such that its external behavior perfectly resembled a human. It morphed into a humanoid shape, did human things… but you’d still say it’s not human. It’s all the intelligence, none of the humanity.

Would you consider my humanoid slime mold a “person?” You’re free to, but I think it’s arguable. My dog really connects with me… I don’t believe my slime mold would ever really connect with me.

permalink

fedilink

source

parent

[–] CatAssTrophy@safest.space 1 point 2 months ago

The difference of scale is in the grey area of the difference of scale of quantum and classical mechanics, though. Conciousness very much could be something that depends on the emergent properties of quantum mechanics and doesn't reach classical mechanics.

permalink

fedilink

source

parent