There is a fundamental assumption made about the ability of AI here that I believe is wrong.
It assumes that the outputs are lacking because of a limit of ability.
I think there is a strong case to make that many of their limitations come from them doing what we have told them to do. Hallucinations are the stand out example of this. If you train it to give answers to questions, it will answer questions, but it might have to make up the answer to do so. This isn't not knowing that it does not know. This is doing the task given to it regardless of whether it knows or not.
If you were given the task of writing the script for a TV show with the criteria that it not offend any people whatsoever. You are told to make something that is as likeable as you can make it without anyone not-liking it at all. The options for what you can do are reduced to something that is okay-ish but rather bland.
That's what AI is giving us. OK but rather bland. It's giving it to us because that's what we've told it we want.
> I think there is a strong case to make that many of their limitations come from them doing what we have told them to do. Hallucinations are the stand out example of this. If you train it to give answers to questions, it will answer questions, but it might have to make up the answer to do so. This isn't not knowing that it does not know. This is doing the task given to it regardless of whether it knows or not.
Are you asserting that an LLM could be NOT trained to answer when it knows it doesn’t know the answer, or if that’s not possible be trained to NOT answer when it knows it doesn’t know the answer?
If so, I would believe your thinking, but for some reason I have not yet seen a single LLM that behaves with that kind of self-knowledge.
It should be trained to answer when it knows the answer, and to state that it does not know the answer when it does not. They might already have a very good understanding of not knowing internally, but are just not trained to express that.
This is not a problem in the ability of the system, it is a problem of how to construct training for such a task.
To provide training examples where it answers it does not know the answer only when it does not know the answer. You need training examples where it says it doesn't know when it does not contain that knowledge, but it provides an answer when it does know the answer.
To create such an example, you need to know in advance what the model knows and what the model does not know. You can't just have a database of facts that it knows, because you also need to count things that it can readily infer.
Any model that can reliably give the sum of any two 10 digit integers should be able to answer so. You can't list every possible number that a model knows how to add. That is just the tiniest subset of the task you would have to do because you have to determine every inferrable fact, not just integers. Adding to the problem is that training on questions like this can add to the knowledge base to the model either from the question itself or by inductively figuring out the answer based upon the combination of the question and the fact that it was not expected to know the answer.
A completely different training system would have to be implemented. There is research on categorising patterns of activations that can determine a form of 'mental state' of a model. A dynamic training approach where the answer that the model is expected-to-give/rewarded-for-giving is partially dependent on the models own state could be achieved through this mechanism.
There has been quite a lot of work in this area. Analysis of activations around hallucinations seems to show that there is some representation of not knowing.
But again things are not quite so simple. Detecting hallucinations might yield representations where it knew the answer but elected to hallucinate anyway because of some other obscure interaction.
Anthropics work on autoencoding activations for analysis has yielded a lot of information about the inner semantic information on models. I haven't seen a lot on bounds of knowledge there, but I wonder if that's something they hold back for competitive advantage.
I don't say this, because I know how, but because I see no reason why we will be unable to crack that problem. If our brains can do it, so will AI one day.
We have various methods of measuring individual intelligence (which are pretty sketchy imo). But do we have any way to measure or quantify the intelligence of the larger structures that mediate our thought? How do you measure the intelligence of a university, or a business? How much intelligence is contained within a collection of books and papers? To what degree do the tools we use amplify our intelligence?
I see students obsess every day over their SAT scores, which to some is a measure of individual intelligence. But what SAT score would a pair of students working together on a single test get? Or a dozen students working together? Would it be higher or lower? What sort of strategies would maximize their ability to collaborate? What would be the effect of giving/removing access to a calculator on a student's score? Access to scratch paper? Access to textbooks? Access to a dictionary? Access to unlimited time?
If we want to claim to understand intelligence, these are the sort of questions we should be able to answer. Can we?
One measure is the ability to find an optimal solution among many options. From solving an integration problem (option = equation, optimal = correct) to life (option = life choice, optimal = best QOL).
Statistically, an LLM finds a better optimal solution than one individual. But LLM solutions are more similar to each other than individuals’, so among many individuals, a few find a better optimal solution than the LLMs.
Hence why I think we need AI that responds more uniquely. Whether that be fine-tuned local models, one LLM that’s more influenced by its prompt and well-supports extremely long prompts, AI tools for humans, or something else. I’d love to see a local AI with continuous learning that generates complex UX to better interact with the user (more than prompts), but that may be far-fetched…
Substance that can hold State that can allow the substance to produce more of the same state in a dangerous universe full of entropy that wants to remove state-full entities by default - and which is prowled by other state-full entities which predate one another, in such a way that game theory and evolutionary pressure applies.
That is not true. We can detect the presence of a thing by the observation of something it causes. That does not imply we have a good definition of the thing.
At that point, we can only define it as something that causes this observation. And that is not very useful.
We might both agree its a poor definition, but at least it's a poor definition that's observable and neutral. That's useful in regards to this conversation.
Where as the other answer is simply "I can't say, but it's not [this]" and that isn't useful at all. It's simply opinion, which is literally the worst definition around. Personally, I don't define intelligence as "whatever qsera acknowledges as intelligence with no qualifying context"...
I have a two year old that can't count. She can definitely walk around, talk, eat, and do all sorts of other things that I would argue indicate some form of intelligence.
Same for my dog.
Same for entire cultures, including adult humans (anumeric cultures).
Arguably, they're constantly doing calculus, because they can walk, throw, catch, etc - but none of them can count.
---
That said, to the point of my original comment - forcing a definition was absolutely useful, because it allows a more detailed and interesting conversation than "qsera doesn't believe an LLM is intelligent".
I agree that one is a poor definition, though. I also don't really think it rules out LLMs.
but that's exactly the point. Is that intelligent?
We also see robots excel at chess and go, is that not intelligent? Why not? Given that's the mental opposite of the physical examples you're happy to discard (side note: I've done real world robotics development, I'd call walking/catching/throwing considerably more difficult activities than chess/go, and much closer to intelligence).
We see LLMs absolutely blow through the turing test, and that was literally the philosophical "gold standard" for machines that exhibit human-like intelligence for like 70 years.
So I really don't think I'm "getting it all very wrong" - I think this is a fundamental question that you're basically failing to honestly engage with because you've already made up your mind.
huh? In the comment above you said you would argue that it is. So is that not your argument?
>I'd call walking/catching/throwing considerably more difficult activities than chess/go, and much closer to intelligence
This is an even bizarre idea. I get that those things are harder to implement in a robot since it requires extremely fine sensors and motor control. But that making it closer to intelligence? That does not make a lot of sense..
Just because we cannot exactly define intelligence does not mean that you can call any arbitrary thing intelligent!
> This is an even bizarre idea. I get that those things are harder to implement in a robot since it requires extremely fine sensors and motor control. But that making it closer to intelligence? That does not make a lot of sense..
Why? It requires understanding a complex sense of self: where am I? How can I move? how do I fit within the environment? How can I change the environment?
Understanding, reasoning, self-awareness, and planning are all considered core aspects of intelligence, and those are all on display during navigation and locomotion.
I'd suggest those are all much closer to intelligence than the rules of chess. More complicated, too.
> Just because we cannot exactly define intelligence does not mean that you can call any arbitrary thing intelligent!
And just because intelligence appears alien in nature to you, doesn't mean it's not there... An ant only has ~250,000 neurons (far less than an LLM) are ants not intelligent? Not at all?
So if you define intelligence simply as "human", then sure - LLMs aren't that. But I also think that's an uninteresting and banal conclusion.
---
So my core point remains - define intelligent. I think it's surprisingly hard to do in a way that rules out LLMs, but doesn't also rule out large categories of things we do probably agree are intelligent.
Yea no worries, just wanted others to know since when I typed the comment, the submission was quite fresh. It was my hope that it would make less people confused.
An ambiguous metric that's been (ab)used quite a lot by human fundamentalists to try to draw a line between what we can do and what machines can do, in order to feel better about themselves.
I like this version much better because most people don't write books and AI is much better at writing than the average person, probably even a few standard deviations above the average.
Why does it need to be the average? It seems to me more like it models the manifold of human knowledge. However we often query for the average, because that is often good enough and gives us quick results, but there is nothing fundamentally preventing us from sending AI into the deep end of under-explored territory and perhaps coming back with something new. It is ultimately the exploration vs exploitation trade off.
> but there is nothing fundamentally preventing us from sending AI into the deep end of under-explored territory and perhaps coming back with something new
What's stopping us is that AI works by manipulating tokens and language but has no connection to reality as it exists. Einstein famously conceived of special relativity by imagining what it would be like to fly alongside a light wave[1]. This is a process that integrates spatial reasoning and imagination informed by living in the real universe where you can see objects moving or waves propagating in a pond. The language only comes later as a means of communicating these intuitions to others.
I wonder how much knowledge can be decoupled from experience, if at all.
If I read thousands of books that explain the details of another civilization in another galaxy, very thoroughly and consistently, but it it just happens to be all made up - did I gain knowledge? More importantly, does what I have in my brain now flip from being fiction to being knowledge if that civilization flipped from not existing to existing? How so, if nothing in my brain, or how I live out the rest of my life, changes in the least, if not a single atom in this galaxy changes (let's ignore that gravity has infinite reach and all that, for the sake of argument)?
If yes, how? What in your definition of knowledge makes that possible?
It's an interesting analogy you're making because... this is the lived reality of a lot of people that are interested in fictional worldbuilding / stories. And it flips to being real in the film Galaxy Quest.
> If I read thousands of books that explain the details of another civilization in another galaxy, very thoroughly and consistently, but it it just happens to be all made up - did I gain knowledge?
sounds a lot like math - made up entities that very thoroughly and consistently fit together.
Besides "secret" knowledge like the know-how at jobs, there's things like unwritten social etiquette (especially as it varies from place to place) or interfacing with physical world – reading about chopping tomatoes is different from experience acquired by actually chopping tomatoes.
It isn't. I constantly have access to non-public information, like the life of my peers and corporate secrets. Is it useful or essential or even desirable for LLM products? Hardly not, but it exists.
Edit: for "not in the training data" yes, humans generally can't know what they can't know.
Not quite. Large amounts of data going into these models has already been curated, otherwise you would get a tremendous amount of wrong answers for even the most basic questions.
It still produces wrong answers regardless though, not because of the training data but because of just... intrinsics. The question is what an acceptable error rate is, how severe those errors are, and whether a human would make comparable errors.
But this debate has parallels with self-driving cars; even if the numbers say that self-driving cars are not perfect but safer than human drivers, anything but perfection will be considered broken or outright illegal.
Not to be too hippy/dippy, but it's only the average of all human knowledge expressible in language. There is plenty of knowledge not expressible this way - for example, the sequence of muscle contraction orchestrated by your brain that allows you to walk. Likewise with feelings like love, pride, etc. There are words for these things, but they're merely labels on an experience that almost all humans know but the specifics of which can't be written down using text.
That is not true. AI can synthesize information, which is the essence of intelligence. And since it knows more than everyone, it is more intelligent too. What they lack is the ability to create information.
Human intelligence is fundamentally motivated by fear and desire, whereas AI operates on an entirely different paradigm. AI lacks human embodiment, and it lacks the political landscapes born out of complex social relationships. Can we truly equate AI's 'intelligence' with what humans call intelligence? Should we even be calling its functionality 'intelligence' at all?
The author argues that overreliance on AI will degrade the overall intelligence of human society, creating a negative feedback loop where future models train on increasingly degraded human data. I agree with this perspective to some extent. However, to definitively claim that human intelligence will only decline is overly simplistic. Rather, we might be about to witness a different facet—or the flip side—of what we have traditionally defined as intelligence.
Socrates once argued that the invention of writing would degrade the essence of human thought and memory. It is true that our capacity for raw memorization declined, but the act of recording enabled knowledge to be transmitted across generations. Couldn't LLMs represent a similar evolutionary trajectory?
It is undeniably true that LLMs atrophy certain cognitive muscles. However, I believe they catalyze development in other areas. In modern society, human discovery and knowledge are effectively monopolized by specific cliques. Without access to prestigious Western journals or incumbent tech giants, the barrier to entry is immense. The open-source community is no exception. For non-native English speakers, breaking into the open-source culture to access shared knowledge is notoriously difficult. But now, by spending a few dollars on an LLM, I can access the collective knowledge of that open-source ecosystem, translated seamlessly into my native language.
There is an old adage in the Korean Windows community: 'Linux is open, but it is not free.' And it’s true. To use Linux, you had to memorize arcane commands, and due to the lack of proper Korean documentation, the learning curve was vastly steeper than Windows. That very learning curve acted as a gatekeeping wall. LLMs explicitly dismantle that wall.
But this dismantling is a two-way street, and it exposes a fatal flaw in the author’s reliance on Shumailov’s 'Model Collapse' theory. The author claims AI compresses the tails of the data distribution, erasing minority viewpoints. What this ignores is that LLMs act as a conduit for cognitive diversity from the non-Western periphery. When a developer in South Korea or Brazil uses an LLM to translate their culturally embedded logic and problem-solving approaches into fluent English, they are injecting entirely new cognitive patterns into the global corpus. This does not compress the tails of the distribution; it actively thickens and extends them by capturing the 'social mind' of populations previously locked out of the internet's primary, English-dominated datasets.
Furthermore, LLMs function as a tool to re-evaluate things we've historically taken for granted—especially in areas that are too complexly intertwined, socio-politically loaded, or vast for the human mind to fully map. Take DeepMind's AlphaDev discovering a faster sorting algorithm as an example; it was a breakthrough achieved precisely because it reasoned from an alien, non-human perspective.
Human learning is fundamentally bottlenecked by environment and bias. Anyone who has interacted with academia knows it is riddled with pervasive prejudices and systemic inefficiencies. In South Korea, for instance, there is an entrenched bias that only researchers with US pedigrees are legitimate, and only papers in specific Western journals matter. This prejudice has prematurely killed countless promising research initiatives. It makes you wonder if the metrics we have long held up as 'superior' or 'correct' are actually deeply flawed. Modern society is too complex for the 'lone genius' model; paradigm shifts now require the intertwined research of multiple collectives. Yet, during this process, political interests often cause dominant groups to gatekeep and exclude others, completely regardless of scientific efficiency. In this context, an AI that lacks our inherent socio-political biases and optimizes purely based on probabilities can actually drive true breakthroughs.
Given all this, the absolute claim that AI unconditionally degrades human intelligence feels flawed. I seriously question whether the 'total sum' of human intelligence is actually experiencing a meaningful decline. Before making such claims, we desperately need to define what 'intelligence' actually means in this new context. The fatal flaw in current AI discourse is the complete lack of nuance—there is no middle ground. Everything is framed as a binary: either purely utopian or purely apocalyptic.
Speaking from personal experience, my cognitive muscle for writing raw code has atrophied because of AI. However, as a non-native English speaker, I used to struggle immensely with naming conventions. Now, my variable naming and overall architectural design capabilities have vastly improved. Conversely, I acutely feel my skills in manual memory layout management and granular code implementation degrading. The trade-off point will be wildly different for every individual.
Whenever I read doom-saying articles like the author's, I can't shake the feeling that they are simply projecting their own subjective anxieties and trying to pass them off as a universal conclusion
The decline brought by writing was not only in memorization. If in previous ages to understand something was to study deeply, eventually the definition may shift to having asked ChatGPT about it and skimmed through the response. This is discussed thorougly in Technopoly, which I'm still reading.
You should also consider that besides the effects of AI adoption in highly technical and scholarly people it will also affect a majority of average workers who may be more vulnerable to atrophy than others.
Not only that, but eventually AI will be native and people's perspective and usage will not be affected by previous generation habits. If people hardly bother to write their own emails, comments or essays then how will the AI-native generation approach that?
Although you make very solid points, I've been leaning to think that the AI effect on society will be shaped by the average user, not users such as yourself and the colleagues you observe in which case the doom-saying starts to make better sense.
We are all shaped by the cultural and social environments we come from, so my perspective is based on the context I’ve experienced. That difference itself is what I find interesting. These kinds of cultural and intellectual frictions make me think more deeply.
I also agree that asking good questions requires some level of foundational knowledge. But at the same time, it’s difficult to ask good questions without that foundation. As you mentioned, more experienced users will likely benefit more from AI.
Still, I think humans tend to find ways to adapt to these kinds of challenges, so it’s possible that new forms of education will emerge. For example, ways of learning how to use cognitive tools like LLMs effectively.
That said, I admit this is an optimistic view. In reality, as you pointed out, the gap will likely widen. People from more educated backgrounds will be better positioned to use these tools effectively, while others may fall further behind.
I agree with your perspective, and I appreciate your thoughts. It’s interesting to see so many different viewpoints—these differences are what make discussions like this engaging.
>In 2024, Ilia Shumailov and colleagues published a paper in Nature with a straight-talking title: AI models collapse when trained on recursively generated data.
Of course, the models are not intelligent. Their generated output reflects the statistical average. And in averaging more and more, you lose a lot of information.
I'll say it again: because we do not have any material focus on pragmatic, disagreement structuring effective communications, (people are not taught how to discuss disagreement) not only is our current AI being massively misunderstood, the human population do not have the discrete language skills to even use AI without massive hallucination issues that they are in control, but do not have the language nuanced understanding to, well, understand.
The reason being, when taught how to effectively disagree all these counterfactual concepts that AI loses manifest, they are logically necessary. But if people are not taught how to explore the landscape of ideas, they become "fascists for the common" and literally create the hellscape civilization we are all trapped within.
We do not teach people, students, how to explore ideas as a group in a collective exploratory discussion. Students get impressed, meaning their peers with social status use the opportunity of a group conversation to impress social hierarchy: they ridicule ideas they think are "outsider concepts", and this behavior is not suppressed by the student's instructors, they often condone and participate. This is at the foundation level, of learning not to offer ideas that might be outside of the mainstream. The ideas and concepts are there, but they are held private because their environment includes bullies that suppress non-mainstream perceptions.
This in itself is negative, but the ramifications are profound: the landscape of ideas is never realized by a material percentage of the students. And those who could have contributed worthwhile insights have been taught to not contribute.
> The ideas and concepts are there, but they are held private because their environment includes bullies that suppress non-mainstream perceptions.
You mean like teachers? Teachers bullies kids that answers wrong in class, so if you try to think for yourself rather than answer according to the book the teacher often makes fun of you and the rest of the kids laugh at you for not saying the right thing.
Just making teachers not do that would probably go a long way to changing the culture. Kids feeling fear of being asked a question in class is a strong signal we are doing something wrong.
I mean the students first, and the teachers second. Just ask any of the "not popular" kids why they do not speak up, and notice they are already afraid to say honestly, they've already been damaged by the bullyscape that is human childhood.
I hope I'm wrong about this - but they sounded like some breed of utopian leftist mad that an AI not specifically trained on their drivel doesn't agree with their utopian ideas, and they believe that this is solely responsible for The Rapture/The Revolution (TM)/real communism not happening. The use of euphemisms and incredible lack of awareness of using 'hellscape' to describe the first world is a bit of a give-away because you usually only hear that sort of hyperbole-that-is-unaware-it-is-hyperbole from leftists mad that capitalism exists.
The rise of AI writing has only been matched by superficial articles comprised of ideas salad that evince no deep theoretical or historical understanding. Crappy writing has and always will exist, AI doesn’t change that, it just makes awful writing grammatical.
It has generally reduced the signal-to-noise ratio of writing, but the signal-to-noise ratio has been absolutely terrible for such a long time we’ve all adapted to better signal detection.
Because it made noise look more like signal. Essentially, that is what AI is. A noise2signal generator, but you specify how a signal should look like. Then, it makes the noise into your specifications. If you specify "make it look like good writing", it will look like good writing, but it will still be noise. It won't be good writing.
Syntactical (grammatical) text is signal but maybe not the kind of meaningful signal we all wish it was. Meaning, of course, resides in the reader, in the practice of reading (Wittgenstein etc), not the author. We used to use writing quality as a proxy for quality of meaning (thought), but that no longer works well.
We are already on the cusp of fully automated reasoning, and once we have fully automated reasoning, OpenAI and Anthropic can just dedicate part of their compute towards generating new high quality novel output, which will then be fed as training data during pretraining of subsequent models.
I don't believe that to be possible in general. Because we've already had Millenia of philosophers attempting to make discoveries through sheer reasoning and with the small in the grand scheme of things exception of formal logic failed to do so. Which leads me to a principle: No matter how smart you are, you still need the real world as a reference.
Once again LLMs will have to be bound to a source of entropy or feedback of some sort as a limit. Sure you might be able to throw terawatts of cycles at say music production but without examples of what people already like or test audiences you cannot answer the question of whether it is any good.
Well, yes, that's why the rest of science was invented, no? I did not mean to imply that AI would restrict itself to philoshical thinking and formal logic.
The reality is that we are hitting a plateau that’s taking us as far as transformer architecture will allow us. Until another breakthrough happens, I heavily doubt we’re on the cusp of anything, instead it’s clear that things are levelling out.
Generating new training data from existing data will only generate patterns that already exist in the training data. It might help LLMs to capture it if has not already. But it can never generate novel patterns.
For example, Imagine some kind of neural network architecture that can do OCR. You might be able to generate variations of the letters it already know using some technique and use it to better train the recognition of already known letters.
But it would never be able to generate letters that it does not know.
There is a fundamental assumption made about the ability of AI here that I believe is wrong.
It assumes that the outputs are lacking because of a limit of ability.
I think there is a strong case to make that many of their limitations come from them doing what we have told them to do. Hallucinations are the stand out example of this. If you train it to give answers to questions, it will answer questions, but it might have to make up the answer to do so. This isn't not knowing that it does not know. This is doing the task given to it regardless of whether it knows or not.
If you were given the task of writing the script for a TV show with the criteria that it not offend any people whatsoever. You are told to make something that is as likeable as you can make it without anyone not-liking it at all. The options for what you can do are reduced to something that is okay-ish but rather bland.
That's what AI is giving us. OK but rather bland. It's giving it to us because that's what we've told it we want.
> I think there is a strong case to make that many of their limitations come from them doing what we have told them to do. Hallucinations are the stand out example of this. If you train it to give answers to questions, it will answer questions, but it might have to make up the answer to do so. This isn't not knowing that it does not know. This is doing the task given to it regardless of whether it knows or not.
Are you asserting that an LLM could be NOT trained to answer when it knows it doesn’t know the answer, or if that’s not possible be trained to NOT answer when it knows it doesn’t know the answer?
If so, I would believe your thinking, but for some reason I have not yet seen a single LLM that behaves with that kind of self-knowledge.
It should be trained to answer when it knows the answer, and to state that it does not know the answer when it does not. They might already have a very good understanding of not knowing internally, but are just not trained to express that.
This is not a problem in the ability of the system, it is a problem of how to construct training for such a task.
To provide training examples where it answers it does not know the answer only when it does not know the answer. You need training examples where it says it doesn't know when it does not contain that knowledge, but it provides an answer when it does know the answer.
To create such an example, you need to know in advance what the model knows and what the model does not know. You can't just have a database of facts that it knows, because you also need to count things that it can readily infer.
Any model that can reliably give the sum of any two 10 digit integers should be able to answer so. You can't list every possible number that a model knows how to add. That is just the tiniest subset of the task you would have to do because you have to determine every inferrable fact, not just integers. Adding to the problem is that training on questions like this can add to the knowledge base to the model either from the question itself or by inductively figuring out the answer based upon the combination of the question and the fact that it was not expected to know the answer.
A completely different training system would have to be implemented. There is research on categorising patterns of activations that can determine a form of 'mental state' of a model. A dynamic training approach where the answer that the model is expected-to-give/rewarded-for-giving is partially dependent on the models own state could be achieved through this mechanism.
> It should be trained to answer when it knows the answer, and to state that it does not know the answer when it does not
Do LLMs even have any kind of internal model of what they know or don't know? My understanding is that they don't.
There has been quite a lot of work in this area. Analysis of activations around hallucinations seems to show that there is some representation of not knowing.
There are things like https://arxiv.org/abs/2410.22071v2
But again things are not quite so simple. Detecting hallucinations might yield representations where it knew the answer but elected to hallucinate anyway because of some other obscure interaction.
Anthropics work on autoencoding activations for analysis has yielded a lot of information about the inner semantic information on models. I haven't seen a lot on bounds of knowledge there, but I wonder if that's something they hold back for competitive advantage.
Of course it's possible.
I don't say this, because I know how, but because I see no reason why we will be unable to crack that problem. If our brains can do it, so will AI one day.
Just wondering... What is Intellgience?
We have various methods of measuring individual intelligence (which are pretty sketchy imo). But do we have any way to measure or quantify the intelligence of the larger structures that mediate our thought? How do you measure the intelligence of a university, or a business? How much intelligence is contained within a collection of books and papers? To what degree do the tools we use amplify our intelligence?
I see students obsess every day over their SAT scores, which to some is a measure of individual intelligence. But what SAT score would a pair of students working together on a single test get? Or a dozen students working together? Would it be higher or lower? What sort of strategies would maximize their ability to collaborate? What would be the effect of giving/removing access to a calculator on a student's score? Access to scratch paper? Access to textbooks? Access to a dictionary? Access to unlimited time?
If we want to claim to understand intelligence, these are the sort of questions we should be able to answer. Can we?
Not sure if you remember this chess game. 1 man vs 50000. Quite an amazing outcome.
https://en.wikipedia.org/wiki/Kasparov_versus_the_World
One measure is the ability to find an optimal solution among many options. From solving an integration problem (option = equation, optimal = correct) to life (option = life choice, optimal = best QOL).
Statistically, an LLM finds a better optimal solution than one individual. But LLM solutions are more similar to each other than individuals’, so among many individuals, a few find a better optimal solution than the LLMs.
Hence why I think we need AI that responds more uniquely. Whether that be fine-tuned local models, one LLM that’s more influenced by its prompt and well-supports extremely long prompts, AI tools for humans, or something else. I’d love to see a local AI with continuous learning that generates complex UX to better interact with the user (more than prompts), but that may be far-fetched…
Substance that can hold State that can allow the substance to produce more of the same state in a dangerous universe full of entropy that wants to remove state-full entities by default - and which is prowled by other state-full entities which predate one another, in such a way that game theory and evolutionary pressure applies.
In psychology, there is no definition of intelligence, but it is generally understood as whatever it is that intelligence tests measure.
I don't have a good answer. But I have a good reason to say LLMs are not intelligent.
https://news.ycombinator.com/item?id=47918103
If you can’t define it you cannot say where it is present or not.
That is not true. We can detect the presence of a thing by the observation of something it causes. That does not imply we have a good definition of the thing.
At that point, we can only define it as something that causes this observation. And that is not very useful.
We might both agree its a poor definition, but at least it's a poor definition that's observable and neutral. That's useful in regards to this conversation.
Where as the other answer is simply "I can't say, but it's not [this]" and that isn't useful at all. It's simply opinion, which is literally the worst definition around. Personally, I don't define intelligence as "whatever qsera acknowledges as intelligence with no qualifying context"...
I think it is useless as a definition, so not even a poor one.
But here you go, https://news.ycombinator.com/item?id=47918103
I have a two year old that can't count. She can definitely walk around, talk, eat, and do all sorts of other things that I would argue indicate some form of intelligence.
Same for my dog.
Same for entire cultures, including adult humans (anumeric cultures).
Arguably, they're constantly doing calculus, because they can walk, throw, catch, etc - but none of them can count.
---
That said, to the point of my original comment - forcing a definition was absolutely useful, because it allows a more detailed and interesting conversation than "qsera doesn't believe an LLM is intelligent".
I agree that one is a poor definition, though. I also don't really think it rules out LLMs.
You are not getting it or getting it all very wrong..
Is a robot that can walk, throw and catch is intelligent?
Last day there was a robot that can play ping pong? Is that intelligent?
but that's exactly the point. Is that intelligent?
We also see robots excel at chess and go, is that not intelligent? Why not? Given that's the mental opposite of the physical examples you're happy to discard (side note: I've done real world robotics development, I'd call walking/catching/throwing considerably more difficult activities than chess/go, and much closer to intelligence).
We see LLMs absolutely blow through the turing test, and that was literally the philosophical "gold standard" for machines that exhibit human-like intelligence for like 70 years.
So I really don't think I'm "getting it all very wrong" - I think this is a fundamental question that you're basically failing to honestly engage with because you've already made up your mind.
So again - define intelligence?
>Is that intelligent?
huh? In the comment above you said you would argue that it is. So is that not your argument?
>I'd call walking/catching/throwing considerably more difficult activities than chess/go, and much closer to intelligence
This is an even bizarre idea. I get that those things are harder to implement in a robot since it requires extremely fine sensors and motor control. But that making it closer to intelligence? That does not make a lot of sense..
Just because we cannot exactly define intelligence does not mean that you can call any arbitrary thing intelligent!
> This is an even bizarre idea. I get that those things are harder to implement in a robot since it requires extremely fine sensors and motor control. But that making it closer to intelligence? That does not make a lot of sense..
Why? It requires understanding a complex sense of self: where am I? How can I move? how do I fit within the environment? How can I change the environment?
Understanding, reasoning, self-awareness, and planning are all considered core aspects of intelligence, and those are all on display during navigation and locomotion.
I'd suggest those are all much closer to intelligence than the rules of chess. More complicated, too.
> Just because we cannot exactly define intelligence does not mean that you can call any arbitrary thing intelligent!
And just because intelligence appears alien in nature to you, doesn't mean it's not there... An ant only has ~250,000 neurons (far less than an LLM) are ants not intelligent? Not at all?
So if you define intelligence simply as "human", then sure - LLMs aren't that. But I also think that's an uninteresting and banal conclusion.
---
So my core point remains - define intelligent. I think it's surprisingly hard to do in a way that rules out LLMs, but doesn't also rule out large categories of things we do probably agree are intelligent.
>where am I? How can I move? how do I fit within the environment?
These does not require any general intelligence or understanding and can be expressed as a bunch of fixed rules.
>I think it's surprisingly hard to do in a way that rules out LLMs,
May be. But we sure can say that LLMs are not "reasoning". Which is what I have done in the linked comment.
yeah! Just like gravity, which we all know is, uh, umm, uh, hold on a second,
Wait, I meant light, yeah, photons! It's photons all the way down! And what are photons you ask? shit, no more questions. Got to go.
https://en.wikipedia.org/wiki/Set_(mathematics)
The titel has a typo as the actual article has the title "The Social Edge of Intelligence".
I've corrected the typo now, but I almost let it stand as a testament to my humanity.
Yea no worries, just wanted others to know since when I typed the comment, the submission was quite fresh. It was my hope that it would make less people confused.
I didn't mind that there was a typo.
I assume most readers could decipher that typo, but I recognize your need to be pedantically correct as part of the grand tradition of HN.
The definition I give to my children: It is the ability to make good decisions.
That is a very reasonable definition, but having grown up playing Dungeons & Dragons, making good decisions strikes me as Wisdom, not Intelligence.
Very "intelligent" people can use those smarts to justify or rationalize all kinds of crazy stupid decisions.
An ambiguous metric that's been (ab)used quite a lot by human fundamentalists to try to draw a line between what we can do and what machines can do, in order to feel better about themselves.
Or were you looking for a different definition?
No consensus but a decent definition is: Ability to utilize resources to achieve outcomes
Like kicking a ball to win a soccer match?
Yes, kicking a ball in the right way does require intelligence.
That sounds more like competency, not intelligence, which funnily enough is only loosely correlated to intelligence.
If you gotta ask, you can't afford it.
~ intelligence
Successful selfishness. The ability to make executable plans to tilt the future in your favor.
This feels like a restating of the idea that for any given endeavor AI raises the floor of quality but doesn't push the ceiling.
My reading of the article is that it claims the ceiling is lowered, especially in the longer term.
Generative AI is the average of all human knowledge
More like: the output of genAI are probably the average knowledge of all the human training data.
I like this version much better because most people don't write books and AI is much better at writing than the average person, probably even a few standard deviations above the average.
Why does it need to be the average? It seems to me more like it models the manifold of human knowledge. However we often query for the average, because that is often good enough and gives us quick results, but there is nothing fundamentally preventing us from sending AI into the deep end of under-explored territory and perhaps coming back with something new. It is ultimately the exploration vs exploitation trade off.
> but there is nothing fundamentally preventing us from sending AI into the deep end of under-explored territory and perhaps coming back with something new
What's stopping us is that AI works by manipulating tokens and language but has no connection to reality as it exists. Einstein famously conceived of special relativity by imagining what it would be like to fly alongside a light wave[1]. This is a process that integrates spatial reasoning and imagination informed by living in the real universe where you can see objects moving or waves propagating in a pond. The language only comes later as a means of communicating these intuitions to others.
[1] https://sites.pitt.edu/~jdnorton/Goodies/Chasing_the_light/
While true in some sense, it does have more knowledge than the average person.
I wonder how much knowledge can be decoupled from experience, if at all.
If I read thousands of books that explain the details of another civilization in another galaxy, very thoroughly and consistently, but it it just happens to be all made up - did I gain knowledge? More importantly, does what I have in my brain now flip from being fiction to being knowledge if that civilization flipped from not existing to existing? How so, if nothing in my brain, or how I live out the rest of my life, changes in the least, if not a single atom in this galaxy changes (let's ignore that gravity has infinite reach and all that, for the sake of argument)?
If yes, how? What in your definition of knowledge makes that possible?
It's an interesting analogy you're making because... this is the lived reality of a lot of people that are interested in fictional worldbuilding / stories. And it flips to being real in the film Galaxy Quest.
> If I read thousands of books that explain the details of another civilization in another galaxy, very thoroughly and consistently, but it it just happens to be all made up - did I gain knowledge?
sounds a lot like math - made up entities that very thoroughly and consistently fit together.
It also does not have access to any knowledge that isn't public or written down or even not in their training data.
Isn’t the same true for a human?
Besides "secret" knowledge like the know-how at jobs, there's things like unwritten social etiquette (especially as it varies from place to place) or interfacing with physical world – reading about chopping tomatoes is different from experience acquired by actually chopping tomatoes.
It isn't. I constantly have access to non-public information, like the life of my peers and corporate secrets. Is it useful or essential or even desirable for LLM products? Hardly not, but it exists.
Edit: for "not in the training data" yes, humans generally can't know what they can't know.
> Is it useful or essential or even desirable for LLM products? Hardly not, but it exists.
Its necessary if you want to replace such a worker, at least to have the corporate culture and knowledge in its training set.
With AI, everyone will be average in no time!
Internet started it, hopefully LLMs will finish it.
“Think of how stupid the average person is, and realize half of them are stupider than that.” — George Carlin
Now with LLMs lowering the average through cognitive offloading and skill atrophy, prepare for it to get a whole lot worse.
For some that'll be an upgrade.
Not quite. Large amounts of data going into these models has already been curated, otherwise you would get a tremendous amount of wrong answers for even the most basic questions.
It still produces wrong answers regardless though, not because of the training data but because of just... intrinsics. The question is what an acceptable error rate is, how severe those errors are, and whether a human would make comparable errors.
But this debate has parallels with self-driving cars; even if the numbers say that self-driving cars are not perfect but safer than human drivers, anything but perfection will be considered broken or outright illegal.
It is the average of sub sets of humans, which can be vastly better than just entire population.
Example it can easily do moderately advanced calculus, that is way better than the average human.
Not to be too hippy/dippy, but it's only the average of all human knowledge expressible in language. There is plenty of knowledge not expressible this way - for example, the sequence of muscle contraction orchestrated by your brain that allows you to walk. Likewise with feelings like love, pride, etc. There are words for these things, but they're merely labels on an experience that almost all humans know but the specifics of which can't be written down using text.
That is not true. AI can synthesize information, which is the essence of intelligence. And since it knows more than everyone, it is more intelligent too. What they lack is the ability to create information.
Human intelligence is fundamentally motivated by fear and desire, whereas AI operates on an entirely different paradigm. AI lacks human embodiment, and it lacks the political landscapes born out of complex social relationships. Can we truly equate AI's 'intelligence' with what humans call intelligence? Should we even be calling its functionality 'intelligence' at all?
The author argues that overreliance on AI will degrade the overall intelligence of human society, creating a negative feedback loop where future models train on increasingly degraded human data. I agree with this perspective to some extent. However, to definitively claim that human intelligence will only decline is overly simplistic. Rather, we might be about to witness a different facet—or the flip side—of what we have traditionally defined as intelligence.
Socrates once argued that the invention of writing would degrade the essence of human thought and memory. It is true that our capacity for raw memorization declined, but the act of recording enabled knowledge to be transmitted across generations. Couldn't LLMs represent a similar evolutionary trajectory?
It is undeniably true that LLMs atrophy certain cognitive muscles. However, I believe they catalyze development in other areas. In modern society, human discovery and knowledge are effectively monopolized by specific cliques. Without access to prestigious Western journals or incumbent tech giants, the barrier to entry is immense. The open-source community is no exception. For non-native English speakers, breaking into the open-source culture to access shared knowledge is notoriously difficult. But now, by spending a few dollars on an LLM, I can access the collective knowledge of that open-source ecosystem, translated seamlessly into my native language.
There is an old adage in the Korean Windows community: 'Linux is open, but it is not free.' And it’s true. To use Linux, you had to memorize arcane commands, and due to the lack of proper Korean documentation, the learning curve was vastly steeper than Windows. That very learning curve acted as a gatekeeping wall. LLMs explicitly dismantle that wall.
But this dismantling is a two-way street, and it exposes a fatal flaw in the author’s reliance on Shumailov’s 'Model Collapse' theory. The author claims AI compresses the tails of the data distribution, erasing minority viewpoints. What this ignores is that LLMs act as a conduit for cognitive diversity from the non-Western periphery. When a developer in South Korea or Brazil uses an LLM to translate their culturally embedded logic and problem-solving approaches into fluent English, they are injecting entirely new cognitive patterns into the global corpus. This does not compress the tails of the distribution; it actively thickens and extends them by capturing the 'social mind' of populations previously locked out of the internet's primary, English-dominated datasets.
Furthermore, LLMs function as a tool to re-evaluate things we've historically taken for granted—especially in areas that are too complexly intertwined, socio-politically loaded, or vast for the human mind to fully map. Take DeepMind's AlphaDev discovering a faster sorting algorithm as an example; it was a breakthrough achieved precisely because it reasoned from an alien, non-human perspective.
Human learning is fundamentally bottlenecked by environment and bias. Anyone who has interacted with academia knows it is riddled with pervasive prejudices and systemic inefficiencies. In South Korea, for instance, there is an entrenched bias that only researchers with US pedigrees are legitimate, and only papers in specific Western journals matter. This prejudice has prematurely killed countless promising research initiatives. It makes you wonder if the metrics we have long held up as 'superior' or 'correct' are actually deeply flawed. Modern society is too complex for the 'lone genius' model; paradigm shifts now require the intertwined research of multiple collectives. Yet, during this process, political interests often cause dominant groups to gatekeep and exclude others, completely regardless of scientific efficiency. In this context, an AI that lacks our inherent socio-political biases and optimizes purely based on probabilities can actually drive true breakthroughs.
Given all this, the absolute claim that AI unconditionally degrades human intelligence feels flawed. I seriously question whether the 'total sum' of human intelligence is actually experiencing a meaningful decline. Before making such claims, we desperately need to define what 'intelligence' actually means in this new context. The fatal flaw in current AI discourse is the complete lack of nuance—there is no middle ground. Everything is framed as a binary: either purely utopian or purely apocalyptic.
Speaking from personal experience, my cognitive muscle for writing raw code has atrophied because of AI. However, as a non-native English speaker, I used to struggle immensely with naming conventions. Now, my variable naming and overall architectural design capabilities have vastly improved. Conversely, I acutely feel my skills in manual memory layout management and granular code implementation degrading. The trade-off point will be wildly different for every individual.
Whenever I read doom-saying articles like the author's, I can't shake the feeling that they are simply projecting their own subjective anxieties and trying to pass them off as a universal conclusion
Great thoughts.
The decline brought by writing was not only in memorization. If in previous ages to understand something was to study deeply, eventually the definition may shift to having asked ChatGPT about it and skimmed through the response. This is discussed thorougly in Technopoly, which I'm still reading.
You should also consider that besides the effects of AI adoption in highly technical and scholarly people it will also affect a majority of average workers who may be more vulnerable to atrophy than others.
Not only that, but eventually AI will be native and people's perspective and usage will not be affected by previous generation habits. If people hardly bother to write their own emails, comments or essays then how will the AI-native generation approach that?
Although you make very solid points, I've been leaning to think that the AI effect on society will be shaped by the average user, not users such as yourself and the colleagues you observe in which case the doom-saying starts to make better sense.
I agree with your point.
We are all shaped by the cultural and social environments we come from, so my perspective is based on the context I’ve experienced. That difference itself is what I find interesting. These kinds of cultural and intellectual frictions make me think more deeply.
I also agree that asking good questions requires some level of foundational knowledge. But at the same time, it’s difficult to ask good questions without that foundation. As you mentioned, more experienced users will likely benefit more from AI.
Still, I think humans tend to find ways to adapt to these kinds of challenges, so it’s possible that new forms of education will emerge. For example, ways of learning how to use cognitive tools like LLMs effectively.
That said, I admit this is an optimistic view. In reality, as you pointed out, the gap will likely widen. People from more educated backgrounds will be better positioned to use these tools effectively, while others may fall further behind.
I agree with your perspective, and I appreciate your thoughts. It’s interesting to see so many different viewpoints—these differences are what make discussions like this engaging.
Have a great day.
Hey, the more we think about our information economy/environment as a commons, the better.
I fully expect our future to involve PhD factories where doctorates label AI output for the most competitive rates possible.
The majority of us will have to contend with an information environment that is polluted and overrun.
I’ll argue with that the internet pre social media was the “healthiest” in terms of our digital commons.
>In 2024, Ilia Shumailov and colleagues published a paper in Nature with a straight-talking title: AI models collapse when trained on recursively generated data.
Of course, the models are not intelligent. Their generated output reflects the statistical average. And in averaging more and more, you lose a lot of information.
I'll say it again: because we do not have any material focus on pragmatic, disagreement structuring effective communications, (people are not taught how to discuss disagreement) not only is our current AI being massively misunderstood, the human population do not have the discrete language skills to even use AI without massive hallucination issues that they are in control, but do not have the language nuanced understanding to, well, understand.
The reason being, when taught how to effectively disagree all these counterfactual concepts that AI loses manifest, they are logically necessary. But if people are not taught how to explore the landscape of ideas, they become "fascists for the common" and literally create the hellscape civilization we are all trapped within.
Can you state your point more simply?
We do not teach people, students, how to explore ideas as a group in a collective exploratory discussion. Students get impressed, meaning their peers with social status use the opportunity of a group conversation to impress social hierarchy: they ridicule ideas they think are "outsider concepts", and this behavior is not suppressed by the student's instructors, they often condone and participate. This is at the foundation level, of learning not to offer ideas that might be outside of the mainstream. The ideas and concepts are there, but they are held private because their environment includes bullies that suppress non-mainstream perceptions.
This in itself is negative, but the ramifications are profound: the landscape of ideas is never realized by a material percentage of the students. And those who could have contributed worthwhile insights have been taught to not contribute.
> The ideas and concepts are there, but they are held private because their environment includes bullies that suppress non-mainstream perceptions.
You mean like teachers? Teachers bullies kids that answers wrong in class, so if you try to think for yourself rather than answer according to the book the teacher often makes fun of you and the rest of the kids laugh at you for not saying the right thing.
Just making teachers not do that would probably go a long way to changing the culture. Kids feeling fear of being asked a question in class is a strong signal we are doing something wrong.
I mean the students first, and the teachers second. Just ask any of the "not popular" kids why they do not speak up, and notice they are already afraid to say honestly, they've already been damaged by the bullyscape that is human childhood.
"democracy basically means, government by the people, of the people , for the people... but the people are retarded"
I hope I'm wrong about this - but they sounded like some breed of utopian leftist mad that an AI not specifically trained on their drivel doesn't agree with their utopian ideas, and they believe that this is solely responsible for The Rapture/The Revolution (TM)/real communism not happening. The use of euphemisms and incredible lack of awareness of using 'hellscape' to describe the first world is a bit of a give-away because you usually only hear that sort of hyperbole-that-is-unaware-it-is-hyperbole from leftists mad that capitalism exists.
Well, my use of 'hellscape' is pretty accurate for where we are as a western culture. No "utopian leftist", just over intellectualized.
[dead]
[dead]
The rise of AI writing has only been matched by superficial articles comprised of ideas salad that evince no deep theoretical or historical understanding. Crappy writing has and always will exist, AI doesn’t change that, it just makes awful writing grammatical.
It has generally reduced the signal-to-noise ratio of writing, but the signal-to-noise ratio has been absolutely terrible for such a long time we’ve all adapted to better signal detection.
Because it made noise look more like signal. Essentially, that is what AI is. A noise2signal generator, but you specify how a signal should look like. Then, it makes the noise into your specifications. If you specify "make it look like good writing", it will look like good writing, but it will still be noise. It won't be good writing.
Syntactical (grammatical) text is signal but maybe not the kind of meaningful signal we all wish it was. Meaning, of course, resides in the reader, in the practice of reading (Wittgenstein etc), not the author. We used to use writing quality as a proxy for quality of meaning (thought), but that no longer works well.
We are already on the cusp of fully automated reasoning, and once we have fully automated reasoning, OpenAI and Anthropic can just dedicate part of their compute towards generating new high quality novel output, which will then be fed as training data during pretraining of subsequent models.
I don't believe that to be possible in general. Because we've already had Millenia of philosophers attempting to make discoveries through sheer reasoning and with the small in the grand scheme of things exception of formal logic failed to do so. Which leads me to a principle: No matter how smart you are, you still need the real world as a reference.
Once again LLMs will have to be bound to a source of entropy or feedback of some sort as a limit. Sure you might be able to throw terawatts of cycles at say music production but without examples of what people already like or test audiences you cannot answer the question of whether it is any good.
Well, yes, that's why the rest of science was invented, no? I did not mean to imply that AI would restrict itself to philoshical thinking and formal logic.
It's proven to be possible in narrow areas like Go. There is no entropy or feedback or whatever. It just keeps getting better.
The reality is that we are hitting a plateau that’s taking us as far as transformer architecture will allow us. Until another breakthrough happens, I heavily doubt we’re on the cusp of anything, instead it’s clear that things are levelling out.
That is like saying we can get unlimited data compression by feeding the output of a data compressing program into its own input..
Not it's not like saying that at all.
Yes, that is exactly like it.
Generating new training data from existing data will only generate patterns that already exist in the training data. It might help LLMs to capture it if has not already. But it can never generate novel patterns.
For example, Imagine some kind of neural network architecture that can do OCR. You might be able to generate variations of the letters it already know using some technique and use it to better train the recognition of already known letters.
But it would never be able to generate letters that it does not know.