188
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 27 Dec 2024
188 points (94.8% liked)
Technology
60346 readers
941 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 2 years ago
MODERATORS
Lol. We're as far away from getting to AGI as we were before the whole LLM craze. It's just glorified statistical text prediction, no matter how much data you throw at it, it will still just guess what's the next most likely letter/token based on what's before it, that can't even get it's facts straith without bullshitting.
If we ever get it, it won't be through LLMs.
I hope someone will finally mathematically prove that it's impossible with current algorithms, so we can finally be done with this bullshiting.
There are already a few papers about diminishing returns in LLM.
I just tried Google Gemini and it would not stop making shit up, it was really disappointing.
I'm not sure that not bullshitting should be a strict criterion of AGI if whether or not it's been achieved is gauged by its capacity to mimic human thought
The LLM aren't bullshitting. They can't lie, because they have no concepts at all. To the machine, the words are all just numerical values with no meaning at all.
Just for the sake of playing a stoner epiphany style of devils advocate: how does thst differ from how actual logical arguments are proven? Hell, why stop there. I mean there isn't a single thing in the universe that can't be broken down to a mathematical equation for physics or chemistry? I'm curious as to how different the processes are between a more advanced LLM or AGI model processing data is compares to a severe case savant memorizing libraries of books using their home made mathematical algorithms. I know it's a leap and I could be wrong but I thought I've heard that some of the rainmaker tier of savants actually process every experiences in a mathematical language.
Like I said in the beginning this is straight up bong rips philosophy and haven't looked up any of the shit I brought up.
I will say tho, I genuinely think the whole LLM shit is without a doubt one of the most amazing advances in technology since the internet. With that being said, I also agree that it has a niche where it will be isolated to being useful under. The problem is that everyone and their slutty mother investing in LLMs are using them for everything they are not useful for and we won't see any effective use of an AI services until all the current idiots realize they poured hundreds of millions of dollars into something that can't out perform any more independently than a 3 year old.
First of all, I'm about to give the extreme dumbed down explanation, but there are actual academics covering this topic right now usually using keywords like AI "emergent behavior" and "overfitting". More specifically about how emergent behavior doesn't really exist in certain model archetypes and that overfitting increases accuracy but effectively makes it more robotic and useless. There are also studies of how humans think.
Anyways, human's don't assign numerical values to words and phrases for the purpose of making a statistical model of a response to a statistical model input.
Humans suck at math.
Humans store data in a much messier unorganized way, and retrieve it by tracing stacks of related concepts back to the root, or fail to memorize data altogether. The values are incredibly diverse and have many attributes to them. Humans do not hallucinate entire documentations up or describe company policies that don't exist to customers, because we understand the branching complexity and nuance to each individual word and phrase. For a human to describe procedures or creatures that do not exist we would have to be lying for some perceived benefit such as entertainment, unlike an LLM which meant that shit it said but just doesn't know any better. Just doesn't know, period.
Maybe an LLM could approach that at some scale if each word had it's own model with massive amounts more data, but given their diminishing returns displayed so far as we feed in more and more processing power that would take more money and electricity than has ever existed on earth. In fact, that aligns pretty well with OpenAI's statement that it could make an AGI if it had Trillions of Dollars to spend and years to spend it. (They're probably underestimating the costs by magnitudes).
Hey, would you have a reference for this? I'd love to read it. Does it apply to deep neural nets? And/or recurrent NNs?
There is this 2023 study from Stanford which states AI likely do not have emergent abilities LINK
And there is this 2020 study by.... OpenAI... which states the error rate is predictable based on 3 factors, that AI cannot cross below the line or approach 0 error rate without exponentially increasing costs several iterations beyond current models, lending to the idea that they're predictable to a fault LINK
There is another paper by DeepMind in 2022 that comes to the conclusion that even at infinite scales it can never approach below 1.69 irreducable error LINK
This all lends to the idea that AI lacks the same Emergent behavior in Human Language.
Thank you!
They did! Here's a paper that proves basically that:
van Rooij, I., Guest, O., Adolfi, F. et al. Reclaiming AI as a Theoretical Tool for Cognitive Science. Comput Brain Behav 7, 616–636 (2024). https://doi.org/10.1007/s42113-024-00217-5
Basically it formalizes the proof that any black box algorithm that is trained on a finite universe of human outputs to prompts, and capable of taking in any finite input and puts out an output that seems plausibly human-like, is an NP-hard problem. And NP-hard problems of that scale are intractable, and can't be solved using the resources available in the universe, even with perfect/idealized algorithms that haven't yet been invented.
This isn't a proof that AI is impossible, just that the method to develop an AI will need more than just inferential learning from training data.
Thank you, it was an interesting read.
Unfortunately, as I was looking more into it, I've stumbled upon a paper that points out some key problems with the proof. I haven't looked into it more and tbh my expertise in formal math ends at vague memories from CS degree almost 10 years ago, but the points do seem to make sense.
https://arxiv.org/html/2411.06498v1
Doesn't that just say that AI will never be cheap? You can still brute force it, which is more or less how back propagation works.
I don't think "intelligence" needs to have a perfect "solution", it just needs to do things well enough to be useful. Which is how human intelligence developed, evolutionarily - it's absolutely not optimal.
Intractable problems of that scale can't be brute forced because the brute force solution can't be run within the time scale of the universe, using the resources of the universe. If we're talking about maintaining all the computing power of humanity towards a solution and hoping to solve it before the sun expands to cover the earth in about 7.5 billion years, then it's not a real solution.
Yeah, maybe you're right. I don't known where the threshold is.
I wonder if the current computational feasibility will cap out improvment of current generation LLMs soon?
The only text predictor I want in my life is T9
Roger Penrose wrote a whole book on the topic in 1989. https://www.goodreads.com/book/show/179744.The_Emperor_s_New_Mind
His points are well thought out and argued, but my essential takeaway is that a series of switches is not ever going to create a sentient being. The idea is absurd to me, but for the people that disagree? They have no proof, just a religious furver, a fanaticism. Simply stated, they want to believe.
All this AI of today is the AI of the 1980s, just with more transistors than we could fathom back then, but the ideas are the same. After the massive surge from our technology finally catching up with 40-60 year old concepts and algorithms, most everything has been just adding much more data, generalizing models, and other tweaks.
What is a problem is the complete lack of scalability and massive energy consumption. Are we supposed to be drying our clothes at a specific our of the night, and join smart grids to reduce peak air conditioning, to scorn bitcoin because it uses too much electricity, but for an AI that generates images of people with 6 fingers and other mangled appendages, that bullshit anything it doesn't know, for that we need to build nuclear power plants everywhere. It's sickening really.
So no AGI anytime soon, but I am sure Altman has defined it as anything that can make his net worth 1 billion or more, no matter what he has to say or do.
Until you can see the human soul under a microscope, we can't make rocks into people.
Is the goal to create a sentient being, or to create something that seems sentient? How would you even tell the difference (assuming it could pass any test a normal human could)?
Powering off a pile of switches is turning it off. Powering off a sentient being is killing it. Not to mention a million other issues it raises.
What is your brain doing if not statistical text prediction?
The show Westworld portrayed it pretty good. The idea of jumping from text prediction to conscience doesn't seem that unlikely. It's basically text prediction on a loop with some exterior inputs to interact.
How to tell me you're stuck in your head terminally online without telling me you're stuck in your head terminally online.
But have something more to read.
Why being so rude?
Did you actually read the article or just googled until you find something that reinforced your prestablished opinion to use as a weapon against a person that you don't even know?
I will actually read it. Probably the only one of us two who would do that.
If it's convincing I may change my mind. I'm not a radical, like many other people are, and my opinions are subject to change.
Funny to me how defensive you got so quick, accusing of not reading the linked paper before even reading it yourself.
The reason OP was so rude is that your very premise of "what is the brain doing if not statistical text prediction" is completely wrong and you don't even consider it could be. You cite a TV show as a source of how it might be. Your concept of what artificial intelligence is comes from media and not science, and is not founded in reality.
The brain uses words to describe thoughts, the words are not actually the thoughts themselves.
https://advances.massgeneral.org/neuro/journal.aspx?id=1096
Think about small children who haven't learned language yet, do those brains still do "stastical text prediction" despite not having words to predict?
What about dogs and cats and other "less intelligent" creatures, they don't use any words but we still can teach them to understand ideas. You don't need to utter a single word, not even a sound, to train a dog to sit. Are they doing "statistical text prediction" ?
So agi is statistical emotion prediction we then assign logic to
Read other replies I gave on your same subject. I don't want to repeat myself.
But words DO define thoughts, and I gave several examples. Some of them with kids. Precisely in kids you can see how language precedes actual thoughts. I will repeat myself a little here but you can clearly see how kids repeat a lot phrases that they just dont understand, just because their beautiful plastic brains heard the same phrase in the same context.
Dogs and cats are not proven to be conscious as a human being is. Precisely due the lack of an articulate language. Or maybe not just language but articulated thoughts. I think there may be a trend to humanize animals, mostly to give them more rights (even I think that a dog doesn't need to have a intelligent consciousness for it to be bad to hit a dog), but I'm highly doubtful that dogs could develop a chain of thoughts that affects itself without external inputs, that seems a pretty important part of the consciousness experience.
The article you link is highly irrelevant (did you read it? Because I am also accusing you of not reading it as just being result of a quick google to try point your point using a fallacy of authority). The fact that spoken words are created by the brain (duh! Obviously, I don't even know why the how the brain creates an articulated spoken word is even relevant here) does not imply that the brain does not also take form due to the words that it learns.
Giving an easier to understand example. For a classical printing press to print books, the words of those books needed to be loaded before in the press. And the press will only be able to print the letters that had been loaded into it.
the user I replied not also had read the article but they kindly summarize it to me. I will still read it. But its arguments on the impossibility of current LLM architectures to create consciousness are actually pretty good, and had actually put me on the way of being convinced of that. At least by the limitations spoken by the article.
Your analogy to mechanical systems are exactly where the breakdown to comparison with the human brain occurs, our brains are not like that, we don't only have the blocks of text loaded into us, sure we only learn what we get exposed to but that doesn't mean we can't think of things we haven't learned about.
The article I linked talks about the separation between the formation of thoughts and those thoughts being translated into words for linguistics.
The fact that you "don’t even know why the how the brain creates an articulated spoken word is even relevant here" speaks volumes to how much you understand the human brain, particularly in the context of artificial intelligence actually understanding the words it generates and the implications of thoughts behind the words and not just guessing which word comes next based on other words, the meanings of which are irrelevant.
I can listen to a song long enough to learn the words, that doesn't mean I know what the song is about.
Can you think of a colour have you never seen? Could you imagine the colour green if you had never seen it?
The creative process is more modification than creation. taking some inputs, mixing them with other inputs and having an output that has parts of all out inputs, does it sound familiar? But without those input seems impossible to create an output.
And thus the importance of language in an actual intelligent consciousness. Without language the brain could only do direct modifications of the natural inputs, of external inputs. But with language the brain can take an external input, then transform it into a "language output" and immediately take that "language output" and read it as an input, process it, and go on. I think that's the core concept that makes humans different from any other species, this middle thing that we can use to dialogue with ourselves and push our minds further. Not every human may have a constant inner monologue, but every human is capable to talking to themself, and will probably do when making a decision. Without language (language could take many forms, not just spoken language, but the more complex feels like it would be the better) I don't know how this self influence process could take place.
It's a basic argument of generative complexity, I found the article some years ago while trying to find an earlier one (I don't think by the same author) that argued along the same complexity lines, essentially saying that if we worked like AI folks think we do we'd need to be so and so much trillion parameters and our brains would be the size of planets. That article talked about the need for context switching in generating (we don't have access to our cooking skills while playing sportsball), this article talks about the necessity to be able to learn how to learn. Not just at the "adjust learning rate" level, but mechanisms that change the resulting coding, thereby creating different such contexts, or at least that's where I see the connection between those two. In essence: To get to AGI we need AIs which can develop their own topology.
As to "rudeness": Make sure to never visit the Netherlands. Usually how this goes is that I link the article and the AI faithful I pointed it out to goes on a denial spree... because if they a) are actually into the topic, not just bystanders and b) did not have some psychological need to believe (including "my retirement savings are in AI stock") they c) would've come across the general argument themselves during their technological research. Or came up with it themselves, I've also seen examples of that: If you have a good intuition about complexity (and many programmers do) it's not unlikely a shower thought to have. Not as fleshed out as in the article, of course.
That seems a very reasonable approach on the impossibility to achieve AGI with current models..
The first concept I was already kind of thinking about. Current LLM are incredibly inefficient. And it seems to be some theoretical barrier in efficiency that no model has been able to surpass. Giving that same answer that with the current model they would probably need to have trillions of parameters just to stop hallucinating. Not to say that to give them the ability to do more things that just answering question. As this supposedly AGI, even if only worked with word, it would need to be able to do more "types of conversations" that just being the answerer in a question-answer dialog.
But I had not thought of the need of repurpose the same are of the brain (biological or artificial) for doing different task on the go, if I have understood correctly. And it seems pretty clear that current models are unable to do that.
Though I still think that an intelligent consciousness could emerge from a loop of generative "thoughts", the most important of those probably being language.
Getting a little poetical. I don't think that the phrase is "I think therefore I am", but "I can think 'I think therefore I am' therefore I am".
Does a dog have the Buddha nature?
...meaning to say: Just because you happen to have the habit of identifying your consciousness with language (that's TBH where the "stuck in your head" thing came from) doesn't mean that language is necessary, or even a component of, consciousness, instead of merely an object of consciousness. And neither is consciousness necessary to do many things, e.g. I'm perfectly able to stop at a pedestrian light while lost in thought.
What Descartes actually was getting at is "I can't doubt that I doubt, therefore, at least my doubt exists". He had a bit of an existential crisis. Unsolicited Advice has a video about it.
It may be because of the habit.
But when I think of how to define a consciousness and divert it from instinct or reactiveness (like stopping at a red light). I think that something that makes a conscience a conscience must be that a conscience is able to modify itself without external influence.
A dog may be able to fully react and learn how to react with the exterior. But can it modify itself the way human brain can?
A human being can sit alone in a room and start processing information by itself in a loop and completely change that flux of information onto something different, even changing the brain in the process.
For this to happen I think some form of language, some form of "speak to yourself" is needed. Some way for the brain to generate an output that can be immediately be taken as input.
At this point of course this is far more philosophical than technical. And maybe even semantics of "what is a conscience".
Human brains also do processing of audio, video, self learning, feelings, and many more that are definitely not statistical text. There are even people without "inner monologue" that function just fine
Some research does use LLM in combination with other AI to get better results overall, but purely LLM isn't going to work.
Yep, of course. We do more things.
But language is a big thing in the human intelligence and consciousness.
I don't know, and I would assume that anyone really know. But people without internal monologue I have a feeling that they have it but they are not aware of it. Or maybe they talk so much that all the monologue is external.
Interesting you focus on language. Because that's exactly what LLMs cannot understand. There's no LLM that actually has a concept of the meaning of words. Here's an excellent essay illustrating my point.
But this "concepts" of things are built on the relation and iteration of this concepts with our brain.
A baby doesn't born knowing that a table is a table. But they see a table, their parents say the word table, and they end up imprinting that what they have to say when they see that thing is the word table. That then they can relation with other things they know. I've watched some kids grow and learn how to talk lately and it's pretty evident how repetition precedes understanding. Many kids will just repeat words that they parents said in certain situation when they happen to be in the same situation. It's pretty obvious with small kids. But it's a behavior you can also see a lot with adults, just repeating something they heard once they see that particular words fit the context
Also it's interesting that language can actually influence the way concepts are constructed in the brain. For instance ancient greeks saw blue and green as the same colour, because they did only have one word for both colours.
But an LLM isn't actually language. It's numbers that represent tokens that build words. It doesn't have the concept of a table, just the numerical weighting of other tokens related to "tab" & "le".
I don't know how to tell you this. But your brain does not have words imprinted in it...
The concept of this is, funnily enough, something that is being studied that derived from language. For instance ancient greeks did not distinguish between green and blue, as both colours had the same word.
Um, something wrong with your brain buddy? Because that's definitely not at all how mine works.
Then why you just expressed in a statistical prediction manner?
You saw other people using that kind of language while being derogatory to someone they don't like on the internet. You saw yourself in the same context and your brain statistically chose to use the same set of words that has been seen the most in this particular context. Literally chatgtp could have been given me your exact same answer if it would have been trained in your same echo chamber.
Have you ever debated with someone from the polar opposite political spectrum and complain that "they just repeat the same propaganda"? Doesn't it sound like statistical predictions to you? Very simple those, there can be more complex one, but our simplest ways are the ones that define what are the basics of what we are made of.
If you at least would have given me a more complex expression you may had an argument (as humans our process could far more complex an hide a little what we seem to actually be doing it). But in instances like this one, when one person (you) responded with a so obvious statistical prediction on what is needed to be said in a particular complex just made my case. thanks.
ok buddy