I’m not sure what that article is supposed to prove. They are using sone computational language and focusing physical responses to visual stimuli but I don’t think it shows “neural computations” as being equivalent to the kinds of computations done by a TM.
One of the chief functions of our brains is to predict the next thing that going to happen, where it's the images we see or the words we hear. That's not very different from genML predicting the next word.
Why do people keep saying this, very obviously human beings are not LLMs.
I'm not even saying that human beings aren't just neural networks. I'm not even saying that an LLM couldn't be considered intelligent theoretically. I'm not even saying that human beings don't learn through predictions. Those are all arguments that people can have. But human beings are obviously not LLMs.
Human beings learn language years into their childhood. It is extremely obvious that we are not text engines that develop internal reason through the processing of text. Children form internal models of the world before they learn how to talk and before they understand what their parents are saying, and it is based on those internal models and on interactions with non-text inputs that their brains develop language models on top of their internal models.
LLMs invert that process. They form language models, and when the language models get big enough and get refined enough, some degree of internal world-modeling results (in theory, we don't really understand what exactly LLMs are doing internally).
Furthermore, even when humans do develop language models, human language models are based on a kind of cooperative "language game" where we predict not what word is most likely to appear next in a sequence, but instead how other people will react and change our separately observed world based on what we say to them. In other words, human beings learn language as tool to manipulate the world, not as an end in and of itself. It's more accurate to say that human language is an emergent system that results from human beings developing other predictive models rather than to say that language is something we learn just by predicting text tokens. We predict the effects and implications of those text tokens, we don't predict the tokens in isolation of the rest of the world.
Not a dig against LLMs, but I wonder if the people making these claims have ever seen an infant before. Your kid doesn't learn how shapes work based on textual context clues, it learns how shapes work by looking at shapes, and then separately it forms a language model that helps it translate that experience/knowledge into a form that other people can understand.
"But we both just predict things" -- prediction subjects matter. Again, nothing against LLMs, but predicting text output is very different from the types of predictions infants make, and those differences have practical consequences. It is a genuinely useful way of thinking about LLMs to understand that they are not trying to predict "correctness" or to influence the world (minor exceptions for alignment training aside), they are trying to predict text sequences. The task that a model is trained on matters, it's not an implementation detail that can just be discarded.
This is obvious, but for some reason some people want to believe that magically a conceptual framework emerges because animal intelligence has to be something like that anyway.
I don't know how animal intelligence works, I just notice when it understands, and these programs don't. Why should they? They're paraphrasing machines, they have no problem contradicting themselves, they can't define adjectives really, they'll give you synonyms. Again, it's all they have, why should they produce anything else?
It's very impressive, but when I read claims of it being akin to human intelligence that's kind of sad to be honest.
> They're paraphrasing machines, they have no problem contradicting themselves, they can't define adjectives really, they'll give you synonyms. Again, it's all they have, why should they produce anything else?
It can certainly do more than paraphrasing. And re: the contradicting nature, humans do that quite often.
Not sure what you mean by "can't define adjectives"
It isn’t that simple. There’s a part of it that generates text but it does some things that don’t match the description. It works with embeddings (it can translate very well) and it can be ‘programmed’ (ie prompted) to generate text following rules (eg. concise or verbose, table or JSON) but the text generated contains same information regardless of representation. What really happens within those billions of parameters? Did it learn to model certain tasks? How many parameters are needed to encode a NAND gate using an LLM? Etc.
I’m afraid once you hook up a logic tool like Z3 and teach the llm to use it properly (kind of like bing tries to search) you’ll get something like an idiot savant. Not good. Especially bad once you give it access to the internet and a malicious human.
The Sapir-Wharf hypothesis (that human thought reduces to languages) has been consistently refuted again and again. Language is very clearly just a facade over thought, and not thought itself. At least in human minds.
Yes but a human being stuck behind a keyboard certainly has their thoughts reduced to language by necessity. The argument that an AI can’t be thinking because it’s producing language is just as silly, that’s the point
Thank you, a view of consciousness based in reality, not with a bleary-eyed religious or mystical outlook.
Something which oddly seems to be in shorter supply than I'd imagine in this forum.
There's lots of fingers-in-ears denial about what these models say about the (non special) nature of human cognition.
Odd when it seems like common sense, even pre-LLM, that our brains do some cool stuff, but it's all just probabilistic sparks following reinforcement too.
You are hand-waving just as much of not more than those you claim are in denial. What is a “probabilistic spark”? There seems to be something special in human cognition because it is clearly very different unless you think humans are organisms for which the laws of physics don’t apply.
By probabilistic spark I was referring to the firing of neurons in a network.
There "seems to be" something special? Maybe from the perspective of the sensing organ, yes.
However consider that an EEG can measure brain decision impulse before you're consciously aware of making a decision. You then retrospectively frame it as self awareness after the fact to make sense of cause and effect.
Human self awareness and consciousness is just an odd side effect of the fact you are the machine doing the thinking. It seems special to you. There's no evidence that it is, and in fact, given crows, dogs, dolphins and so on show similar (but diminished reasoning) while it may be true we have some unique capability ... unless you want to define "special" I'm going to read "mystical" where you said "special".
Unfortunately we still don't know how it all began, before the big bang etc.
I hope we get to know everything during our lifetimes, or we reach immortality so we have time to get to know everything. This feels honestly like a timeline where there's potential for it.
It feels a bit pointless to have been lived and not knowing what's behind all that.
But what’s going on inside an LLM neural network isn’t ‘language’ - it is ‘language ingestion, processing and generation’. It’s happening in the form of a bunch of floating point numbers, not mechanical operations on tokens.
Who’s to say that in among that processing, there isn’t also ‘reasoning’ or ‘thinking’ going on. Over the top of which the output language is just a façade?
To me, all I know of you is words on the screen, which is the point the parent comment was making. How do we know that we’re both humans when the only means we have to communicate thoughts with each other is through written words?
