I think about this but like, I’m not sure if it’s a stupid question so I’ll put it here.
You must log in or # to comment.
Yes. Humans are fundamentally subject to the same laws of physics that machines are. However, it took about three billion years from the first life to the first sapient life, so I wouldn’t count on machines getting there anytime soon.
Short answer? Yes.
Consciousness arises out of complexity. The awareness that we have is Jill derived from chemical reactions in our brains triggering things in response to stimuli.
Our responses are complex because the system that generates them is complex.
It’s an arrogant assumption that humans are special, that there’s something that makes us more than just animals, that a machine whose systems are complex enough could never be like us.
That said…. As they stand now, LLMs will never be Artificial General Intelligence.They barely rate Artificial Narrow Intelligence. (They’re pattern recognition systems with a huge hash of data to draw on. It’s not particularly complex as far as AGI goes.)
This is a really good question but far above Lemmy’s pay-grade man. Neuroscientists are very actively researching and doing debates on this topic (consciousness in AI)
The main problem is that we don’t even have a consensus, fool-proof theory of consciousness… We can barely tell whether a lot of living creatures are conscious or not (think octopus with their distributed intelligence, plants with their chemical signaling, …), and if so, when in their development stage does consciousness arise. Machines are one step removed from even that; even if we give AI models the benefit of the doubt and say they can develop consciousness, how in the world are we supposed to measure it
I found this highly-cited Nature Reviews Neuroscience paper (Seth, A.K., Bayne, T. Theories of consciousness. Nat Rev Neurosci 23, 439–452 (2022). https://doi.org/10.1038/s41583-022-00587-4) (try Internet Archive if you can’t access it) that might be of interest… If you follow the citation trails, there are a ton of new articles on debating consciousness in AI. Like this arxiv article on consciousness in AI
If AI (LLM) have a consciousness, would that not mean that they only think when they receive input, because they are not computing otherwise?
I guess so, if it is true that would mean current AI companies did some serious ethics fuck-ups…
I’m giving current-generation AI the benefit of the doubt though. I personally don’t think the current transformer-based LLM designs can achieve consciousness (although I think machines in general can, just not the current model architectures)
There is also the question if they can perceive time, or have some kind of emotion (if we assume they are conscious). An interesting topic to ponder.
But I agree, I don’t think that current AI will be conscious, but I don’t see it as impossible just very complex/difficult.
Did we?
We have no idea what consciousness is.
Literally. It’s something that we all experience, but there is no scientific definition, measure, or framework to study it.
We assume it is a function of our brains, but we could be wrong. Are intelligent birds, apes, and dolphins conscious? What about lizards? Flatworms? Trees? Yeast?
If you want to go down that rabbit hole, search “what’s it like to be a bat?” or “the hard problem.”
Consciousness or whatever we want to call “it” doesn’t have a a threshold. It’s not like there’s an easy answer for it. It’s not a “you’re outside, step though the door, now you’re inside” kind of thing.
It’s a spectrum. So yeah. That tapeworm is on the spectrum, but very low down. (Probably consistent with artificial narrow intelligence, maybe,)
It’d be great if any part of that were testable.
That’s true. Every living organism could be conscious on some level if you think about it.
If you’d like to go all the way down this rabbit hole there’s panpsychism, which posits that consciousness is a continuous scale with “zero” literally being down in the elementary particles. So even a single molecule is not quite zero on the scale, having an extremely minuscule degree of consciousness to it.
I actually kind of like this because it means we don’t have to try to put that zero point at some other arbitrary location.
Although I think it’s probably not a bad idea to think about it this way, it does create the next unhelpful question of “when is something conscious enough?”, so it kinda dodges the hard question. So it simultaneously is useful and useless to think about it in this way.
I think your question is equally poorly defined regardless of where the zero point is placed. Enough for what? What is this threshold for?
There are often things in nature that have natural “division points” along some particular spectrum, where conditions abruptly change (phase changes in chemicals when they pass through transition temperatures, for example) or where you have natural gaps where you don’t find many examples because of dynamic feedback effects (the gap between planets and dwarf planets when it comes to their ability to clear their orbital neighborhoods, for example). It’s possible that once we can actually measure consciousness and put a number to it we’ll find that there are similar gaps, for whatever reason. That would be a very interesting and useful thing to know about. But that’s putting the cart before the horse, we don’t know how to measure consciousness in a numeric way yet. So any such patterns in the results are purely hypothetical.
Yeah but that’s exactly what I mean when I say that defining consciousness as a scale brings us no closer to the (poorly defined) questions we have about these things. We don’t even know if we should define it as a one-dimensional number.
“When is something conscious?”
“Well it’s a scale so once it has a high enough consciousness number”
“Okay, so at what point in the scale do we consider it conscious?”
“Well that depends on how you want to define ‘conscious’”
So you get stuck with essentially the same question. And I think that expressing it as a scale also runs into the problem of “how do you measure it then?” which basically requires answering the original consciousness-question again.
@bookie @gedaliyah And rivers and crystals are alive, if you define “alive” broadly enough.
What, then, makes a given definition useful? True?
What does it mean to be conscious?
You’re not wrong, and it’s a really interesting question. The definition of life that I learned in science class includes things like: homeostasis, response to stimuli, reproduction, metabolism, cell growth, adaptation, cell signaling, and death. Rivers adapt, respond to stimuli, and can “die”; maybe reproduce, achieve homeostasis, and metabolize; but don’t have cells. Crystals reproduce and maybe metabolize, but not the rest.
That’s a faulty definition, because it doesn’t clearly define viruses as alive or not alive, but it’s still useful, because the things that we can categorize as alive (or sometimes alive, in the case of viruses) under it share a lot of common characteristics that make sense to study together.
Rivers and crystals have some of these characteristics, but so does everything in nature- a star metabolizes and dies, a planet achieves homeostasis and responds to stimuli, and asteroids traveling fast enough can “reproduce” by hitting another stellar object hard enough to break one or both of them into smaller pieces. Including everything in the natural world under the umbrella of life doesn’t help us understand the world around us better according to our current understanding of the universe.
I think it’s impossible for us to determine truth, and the best we can do is an approximate theory that holds up with new information. Even when we know our model is untrue (like how we know classifying species the way we do is generally too rigid, or the fact that there’s no such thing as a fish), we can still gather information better under that model than without it.
Or so we think, maybe we’d have a much clearer understanding of the world if carcinology fell under entomology instead of marine biology.
As for consciousness, that’s even harder to define.
Define consciousness.
It is not a stupid question.
AI is a statistical model. It emulates how we speak and therefore mimics how we think. But there are essential differences currently.
The size of AI-models is way less than that of our brains (comparing nodes to brain cells) by several orders of magnitude. AI models do simulate associative thinking and pattern thinking really well in some ways, but not very well in others. While an AI can have different sections of brain that can go back and forth (Mixture of Experts), they usually only pass raw numbers through the model one way. Our brain afaik is unidirectional with its electrical signals.
I’m saying this because an AI and a brain have fundamental differences we won’t be able to eliminate as of now. Which means a traditional consciousness as we know from humans or animals in some cases is hard to presume.
That said, a lot of experts recently raised alarms because what is classified as a person, when does thinking start and stop, all of these ethical boundaries need more strict definitions law-wise, because we are definitely fishing at the boundaries.
I know Lemmy doesn’t like AI and will highlight these fundamental differences, but the ethical questions raised are indeed part of a growing corpus of science currently.
“AI” consists of statistical prediction models with output. We’ll get Asimov’s psychohistory before we get conscious Gen AI.
LLMs are zip files, not analogous to brains. Model training is a lossy compressiom algorithm.
To answer that question we would first need to understand what consciousness is and how it arises.
Expanding on that: until we can prove that another human being is conscious, it’s highly unlikely that we’ll be able to do so for a machine.
We would need to solve a few dozen philosophical questions before doing anything more than replicating by mimicry. Like some kind of sociopathic entity masking it’s behaviours to seem neurotypical. Probably look like some super jacked Austrian dude who’s going to be back later or maybe a silver surfer in a cop costume, who knows.
Not with our current methods of AI, no.
I completely agree but often get a lot of flak for stating this.
Yeah I wouldn’t expect it right now, but maybe in the future?
Once they start integrating actual brain matter into machines, maybe. But IDK if that would necessarily answer the question of if a machine could develop consciousness, like as in just silicon alone. Is it a machine if it’s just a lab-grown brain attached to electronics? 🤔
I am skeptical that silicon alone would ever develop consciousness, personally.
They do have those computers that use brain cells, I believe it’s called ‘wetware’. They’re trying to see if it’ll be a cheaper alternative to AI. I just feel bad for the brains.
I don’t know if I should feel any more than an insect or other animal. The brain matter isn’t like taking an actual animal’s brain and making it a computer; they’re growing synapses in a lab so even if they were conscious, I feel like it wouldn’t be much different to them as anything else that has always been that thing. Like they wouldn’t have any more of an existential crisis being a computer than we do about living in our own bodies.
Now if they grew sapience that might be something to empathize with.
There are degrees of awareness. Consciousness is a broad part of that, but usually it’s sentience and sapience that are really considered in this question, and they are two different things. Usually sentience is used when it’s actually sapience that’s being discussed.
LLMs as they are aren’t aware. They respond to input, but there’s nothing going on internally to suggest they “know” what’s being asked. The lack of long term context and memory is a huge issue that external tools and techniques try to counter.
But to answer the question, we still don’t know. Before everything got sidetracked onto the LLM ride, it was one of many ideas on how to develop thinking in a computer. I don’t know if other research is still going on and just hidden behind the frenzy, or got cut out to go down this singular path that looked profitable.
I heard there is research to build world models, so I believe they are still trying to build AI beyond LLMs.
There’s a lot broader sorts of AI’s than just LLMs. People are doing all sorts of incredible things with them.
I don’t know why people think LLMs are the most advanced AI we have- very far from it, actually.
Right now most of them are fairly narrow in scope, but they’re still far more complex than LLMs.
The scariest part of this question is that we would have no way of truly knowing.
Not a stupid question. It does not look like at the moment, wouldn’t say it’s impossible though. As we progress with better and better technology I personally think it is an inevitability.
There a plenty of science fiction books and shows that focus on this.
Currently? No. In the future? Who knows?
Hypothetically, if we got to the point that we could simulate the collective synapses of a human brain in real time, that should do it.
Hypothetically, if we got to the point that we could simulate the collective synapses of a human brain in real time, that should do it.
If you simulate fire, does it burn? I don’t think a simulation will do it. I think consciousness is an emergent property, but unlike others here I’m not so sure this property has to emerge from an organic substrate. Organics might just be the cheapest way nature found to do it.
Problem is, we don’t really understand consciousness itself, and that makes it hard to predict where or how it will show up.
If you simulate fire, does it burn?
I hope so, wouldn’t be a good simulation otherwise.
Cute that you think synapses are the defining factor.
Thanks, dick
Coz AI copies us and learns from us, if it did, i would kinda expect it to just act more like us but maybe smarter?
I’ve interacted with quite a few people in my day, and I can’t see how the robots could be much dumber than them.
No
