Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

What is a mathematical structure of conscious experience?

arxiv.org

Post reply on HN
Print view View original

21–30 of 179 posts

Re: What is a mathematical structure of conscious experience?

#21

Post by logicchains »

kilgnad wrote: post #5

>In defining what it means for a mathematical structure to be of conscious experience, our proposal does not answer the question of what this mathematical structure is. Nothing is answered in this article. We simply don't know. This article is up here because nobody is sure whether or not chatGPT is in any way representative of an aspect of our conscious experience. Is chatGPT a mathematical model of consciousness? W…

>Nothing is answered in this article. We simply don't know.

No, we haven't clearly defined consciousness. For a given exact definition of consciousness, we can determine whether something fits that definition or not. If we don't clearly define it, then we're just talking about a feeling, a qualia, and it's not possible for people to reach a consensus because they're not even discussing the exact same thing when they say "consciousness".

Re: What is a mathematical structure of conscious experience?

#22

Post by mxkopy »

goatlover wrote: post #11
mxkopy wrote: post #10

I'm surprised that in all this hype Gödel Escher Bach hasn't seen a proportional comeback. Hofstadter outlines with some formality a very plausible kind of structure of consciousness, namely the "strange loop": a system of symbols or rule-patterns that contains a distinct but isomorphic (or approaching) description of itself.

Does the strange loop explain how sensations like color and pain come from the isomorphic self-description?

The book doesn't really go that much into the hard problem of consciousness, as much of the ideas are orthogonal to that discussion. If Nagel asks, "What is like to be a bat?", Hofstadter asks "If the mind of a bat were a program, what characteristics would it have?"

IMO the hard problem is not posed to be solved. I think of it like this. The world in which we are p-zombies is categorically indistinguishable from the one in which we are not, by an objective perspective. It is not a concept that we should use for an objective definition.

Re: What is a mathematical structure of conscious experience?

#23

Post by mxkopy »

kilgnad wrote: post #12
mxkopy wrote: post #10

I'm surprised that in all this hype Gödel Escher Bach hasn't seen a proportional comeback. Hofstadter outlines with some formality a very plausible kind of structure of consciousness, namely the "strange loop": a system of symbols or rule-patterns that contains a distinct but isomorphic (or approaching) description of itself.

If Hosfadter is correct, then chatGPT already fits that definition. https://www.engraved.blog/building-a-virtual-machine-inside/ Read to the ending where it closes the recursive loop.

In tandem with that article there are also all those findings about GPT training smaller gradient descent models inside itself which is very interesting. That particularly seems rather... strange.

I'm sure Hofstader would dislike people saying "he was right about this!" "he defined this!", though, since that's not what he was setting out to do at all. GEB really is just an exploration, with some of the results being couched in mathematical language as an artifact of his background, not as an attempt to formalize anything.

Re: What is a mathematical structure of conscious experience?

#24

Post by dvt »

kilgnad wrote: post #5

>In defining what it means for a mathematical structure to be of conscious experience, our proposal does not answer the question of what this mathematical structure is. Nothing is answered in this article. We simply don't know. This article is up here because nobody is sure whether or not chatGPT is in any way representative of an aspect of our conscious experience. Is chatGPT a mathematical model of consciousness? W…

> Is chatGPT a mathematical model of consciousness?

Unless we reduce models of consciousness to purely stochastic processes, this is trivially false. And no one seriously reduces consciousness to probabilistic pattern recognition.

> We know it doesn't represent the whole experience...

It doesn't represent any experience. ChatGPT doesn't "experience" things as conscious things do. It's not even comparable to the experience of a simple invertebrate, for example.

Re: What is a mathematical structure of conscious experience?

#25

Post by naasking »

ethanwillis wrote: post #20
kilgnad wrote: post #9

Earlier quoted context omitted.

Completely agreed. There are flaws. But there are times where the output appears that it cannot anything else other then complete understanding of what was given to it as input.

A big lookup table.

A lookup table with compression. Intelligence is a form of compression.

Re: What is a mathematical structure of conscious experience?

#26

Post by Lyapunov_Lover »

dvt wrote: post #24
kilgnad wrote: post #5

>In defining what it means for a mathematical structure to be of conscious experience, our proposal does not answer the question of what this mathematical structure is. Nothing is answered in this article. We simply don't know. This article is up here because nobody is sure whether or not chatGPT is in any way representative of an aspect of our conscious experience. Is chatGPT a mathematical model of consciousness? W…

> Is chatGPT a mathematical model of consciousness? Unless we reduce models of consciousness to purely stochastic processes, this is trivially false. And no one seriously reduces consciousness to probabilistic pattern recognition. > We know it doesn't represent the whole experience... It doesn't represent any experience. ChatGPT doesn't "experience" things as conscious things do. It's not even comparable to the exper…

That's the (hard) problem of qualia. You can't say for sure that ChatGPT doesn't experience anything. And I can't say for sure that you do, either. It's trivial to dismiss it on the basis of "well obviously it's like that"—far more difficult is to say something about why it's not like that.

I think consciousness has to be the product of an algorithmic process. I think this process can be summed up as "prediction". I don't think there's some secret quantum/soul/magic aspect to it. Which is why I think it's plausible that transformer models may capture some aspect of what we refer to as consciousness, though I don't quite buy it at a gut level.

You say that "it's not even comparable to the experience of a simple invertebrate," but that's just you writing down what your gut is telling you. What is the experience of a simple invertebrate like, exactly?

Re: What is a mathematical structure of conscious experience?

#27

Post by kilgnad »

logicchains wrote: post #21
kilgnad wrote: post #5

>In defining what it means for a mathematical structure to be of conscious experience, our proposal does not answer the question of what this mathematical structure is. Nothing is answered in this article. We simply don't know. This article is up here because nobody is sure whether or not chatGPT is in any way representative of an aspect of our conscious experience. Is chatGPT a mathematical model of consciousness? W…

>Nothing is answered in this article. We simply don't know. No, we haven't clearly defined consciousness. For a given exact definition of consciousness, we can determine whether something fits that definition or not. If we don't clearly define it, then we're just talking about a feeling, a qualia, and it's not possible for people to reach a consensus because they're not even discussing the exact same thing when they…

The consistency arises from the fact that even though we don't know the definition, we are in agreement in what is classified as conscious.

We agree that humans are conscious, we agree that rocks are not conscious. This classification is clear and points to a underlying definition we cannot fully articulate. It is more than a qualia or whatever you call it. We know it has exact parameters simply because we are able to perform classification tasks.

The consensus breaks for things like chatGPT. chatGPT straddles some uncertain border that we can't articulate. But in general for the term consciousness we reach consensus by how we classify things as conscious or unconscious.

Re: What is a mathematical structure of conscious experience?

#28

Post by kilgnad »

dvt wrote: post #24
kilgnad wrote: post #5

>In defining what it means for a mathematical structure to be of conscious experience, our proposal does not answer the question of what this mathematical structure is. Nothing is answered in this article. We simply don't know. This article is up here because nobody is sure whether or not chatGPT is in any way representative of an aspect of our conscious experience. Is chatGPT a mathematical model of consciousness? W…

> Is chatGPT a mathematical model of consciousness? Unless we reduce models of consciousness to purely stochastic processes, this is trivially false. And no one seriously reduces consciousness to probabilistic pattern recognition. > We know it doesn't represent the whole experience... It doesn't represent any experience. ChatGPT doesn't "experience" things as conscious things do. It's not even comparable to the exper…

>Unless we reduce models of consciousness to purely stochastic processes, this is trivially false. And no one seriously reduces consciousness to probabilistic pattern recognition.

Most things in this universe have mathematical models that can describe it. We have models for virtually everything. Thus it is reasonable to say such a model also exists for consciousness. To say what you said definitively without even knowing the meaning of the word consciousness is premature. You don't have enough knowledge to say anything about this.

>It doesn't represent any experience. ChatGPT doesn't "experience" things as conscious things do. It's not even comparable to the experience of a simple invertebrate, for example.

Again you don't know what you're talking about. Nobody can define what consciousness is and nobody fully understands the emergent effects of chatGPT or the human brain. To say what you say with conviction is therefore empty. You don't actually know because nobody knows.

Re: What is a mathematical structure of conscious experience?

#29

Post by kilgnad »

mxkopy wrote: post #23
kilgnad wrote: post #12

Earlier quoted context omitted.

If Hosfadter is correct, then chatGPT already fits that definition. https://www.engraved.blog/building-a-virtual-machine-inside/ Read to the ending where it closes the recursive loop.

In tandem with that article there are also all those findings about GPT training smaller gradient descent models inside itself which is very interesting. That particularly seems rather... strange. I'm sure Hofstader would dislike people saying "he was right about this!" "he defined this!", though, since that's not what he was setting out to do at all. GEB really is just an exploration, with some of the results being…

Personally I think Hofstadter was a bit too enamored with recursion. I don't think it's some marker of consciousness. I'm not saying chatGPT isnt an aspect of consciousness (I don't know).

Re: What is a mathematical structure of conscious experience?

#30

Post by dqpb »

mxkopy wrote: post #10

I'm surprised that in all this hype Gödel Escher Bach hasn't seen a proportional comeback. Hofstadter outlines with some formality a very plausible kind of structure of consciousness, namely the "strange loop": a system of symbols or rule-patterns that contains a distinct but isomorphic (or approaching) description of itself.

I've always thought consciousness is basically just feedback (like with microphones). Sensing yourself sensing yourself etc.
Post reply on HN