> That’s also why I see no point in using AI to, say, write an essay, just like I see no point in bringing a forklift to the gym. Sure, it can lift the weights, but I’m not trying to suspend a barbell above the floor for the hell of it. I lift it because I want to become the kind of person who can lift it. Similarly, I write because I want to become the kind of person who can think.
If the motivation structure is there I don’t see an inherent reason for people to refuse cultivating themselves. Going with the gym analogy lay people did not need gyms when physical work was the norm, cultivation was readily accomplished.
If anything there is a competing motivational structure in which people are incentivized not to think but to consume, react, emote etc. Information processing skills of the individual being deliberately eroded/hijacked/bypassed is not a AI thing. The most obvious example is ads. Thinkers are simply not good for business.
Everyone is out here acting like "predicting the next thing" is somehow fundamentally irrelevant to "human thinking" and it is simply not the case.
What does it mean to say that we humans act with intent? It means that we have some expectation or prediction about how our actions will effect the next thing, and choose our actions based on how much we like that effect. The ability to predict is fundamental to our ability to act intentionally.
So in my mind: even if you grant all the AI-naysayer's complaints about how LLMs aren't "actually" thinking, you can still believe that they will end up being a component in a system which actually "does" think.
Are you a stream of words or are your words the “simplistic” projection of your abstract thoughts? I don’t at all discount the importance of language in so many things, but the question that matters is whether statistical models of language can ever “learn” abstract thought, or become part of a system which uses them as a tool.
My personal assessment is that LLMs can do neither.
LLMs and human brains are both just mechanisms. Why would one mechanism a priori be capable of "learning abstract thought", but no others?
If it turns out that LLMs don't model human brains well enough to qualify as "learning abstract thought" the way humans do, some future technology will do so. Human brains aren't magic, special or different.
> Everyone is out here acting like "predicting the next thing" is somehow fundamentally irrelevant to "human thinking" and it is simply not the case.
Nobody is. What people are doing is claiming that "predicting the next thing" does not define the entirety of human thinking, and something that is ONLY predicting the next thing is not, fundamentally, thinking.
Well, yes because thinking soon requires interacting, not just ideating. It's in the dialogue between ideation and interaction that we make our discoveries.
I claim that all of thinking can be reduced to predicting the next thing. Predicting the next thing = thinking in the same way that reading and writing strings of bytes is a universal interface, or every computation can be done by a Turing machine.
I'm an LLMs are being used in workflows they don't make sense in-sayer. And while yes, I can believe that LLMs can be part of a system that actually does think, I believe that to achieve true "thinking", it would likely be a system that is more deterministic in its approach rather than probabilistic.
Especially when modeling acting with intent. The ability to measure against past results and think of new innovative approaches seems like it may come from a system that may model first and then use LLM output. Basically something that has a foundation of tools rather than an LLM using MCP. Perhaps using LLMs to generate a response that humans like to read, but not in them coming up with the answer.
Either way, yes, its possible for a thinking system to use LLMs (and potentially humans piece together sentences in a similar way), but its also possible LLMs will be cast aside and a new approach will be used to create an AGI.
So for me: even if you are an AI-yeasayer, you can still believe that they won't be a component in an AGI.
You can make a separate model for the task, which is based on well chosen features and calibrated from actual data. Then the LLM only needs to generate the arguments to this model (extract those features from messages) and call it like a MCP tool. This external tool can be a simple Sklearn model.
And the big players have built a bunch of workflows which embed many other elements besides just "predictions" into their AI product. Things like web search, to incorporating feedback from code testing, to feeding outputs back into future iterations. Who is to say that one or more of these additions has pushed the ensemble across the threshold and into "real actual thinking."
The near-religious fervor which people insist that "its just prediction" makes me want to respond with some religious allusions of my own:
> Who is this that wrappeth up sentences in unskillful words? Gird up thy loins like a man: I will ask thee, and answer thou me. Where wast thou when I laid up the foundations of the earth? tell me if thou hast understanding. Who hath laid the measures thereof, if thou knowest? or who hath stretched the line upon it?
The point is that (as far as I know) we simply don't know the necessary or sufficient conditions for "thinking" in the first place, let alone "human thinking." Eventually we will most likely arrive at a scientific consensus, but as of right now we don't have the terms nailed down well enough to claim the kind of certainty I see from AI-detractors.
I take a offence in the idea I’m “religiously downplaying LLMs”. I pay top dollar for access to the best models because I want the capabilities to be good / better. Just because I’m documenting my experience it doesn’t mean I have an Anti-ai agenda ? I pay because I find LLMs to be useful. Just not in the way suggested by the marketing teams.
I’m downplaying because I have honestly been burned by these tools when I’ve put trust in their ability to understand anything, provide a novel suggestion or even solve some basic bugs without causing other issues.?
I use all of the things you talk about extremely frequently and again, there is no “thinking” or consideration on display that suggests these things work like us, else why would we be having this conversation if they were ?
> The harms engendered by underestimating LLM capabilities are largely that people won't use the LLMs.
Speculative fiction about superintelligences aside, an obvious harm to underestimating the LLM's capabilities is that we could effectively be enslaving moral agents if we fail to correctly classify them as such.
> The issue is that prediction is "part" of the human thought process, it's not the full story...
Do you have a proof for this?
Surely such a profound claim about human thought process must have a solid proof somewhere? Otherwise who's to say all of human thought process is not just a derivative of "predicting the next thing"?
What would change your mind? It's an exercise in feasibility.
For example, I don't believe in time travel. If someone made me time travel, and made it undeniable that I was transported back to 1508, then I would not be able to argue against it. In fact, no one in such position would.
What is that equivalent for your conviction? There must be something, otherwise, it's just an opinion that can't be changed.
You don't need to present some actual proof or something. Just lay out some ideas that demonstrate that you are being rational about this and not just sucking up to LLM marketing.
When you have a thought, are you "predicting the next thing"—can you confidently classify all mental activity that you experience as "predicting the next thing"?
Language and society constrains the way we use words, but when you speak, are you "predicting"? Science allows human beings to predict various outcomes with varying degrees of success, but much of our experience of the world does not entail predicting things.
How confident are you that the abstractions "search" and "thinking" as applied to the neurological biological machine called the human brain, nervous system, and sensorium and the machine called an LLM are really equatable? On what do you base your confidence in their equivalence?
Does an equivalence of observable behavior imply an ontological equivalence? How does Heisenberg's famous principle complicate this when we consider the role observer's play in founding their own observations? How much of your confidence is based on biased notions rather than direct evidence?
The critics are right to raise these arguments. Companies with a tremendous amount of power are claiming these tools do more than they are actually capable of and they actively mislead consumers in this manner.
> When you have a thought, are you "predicting the next thing"
Yes. This is the core claim of the Free Energy Principle[0], from the most-cited neuroscientist alive. Predictive processing isn't AI hype - it's the dominant theoretical framework in computational neuroscience for ~15 years now.
> much of our experience of the world does not entail predicting things
Introspection isn't evidence about computational architecture. You don't experience your V1 doing edge detection either.
> How confident are you that the abstractions "search" and "thinking"... are really equatable?
This isn't about confidence, it's about whether you're engaging with the actual literature. Active inference[1] argues cognition IS prediction and action in service of minimizing surprise. Disagree if you want, but you're disagreeing with Friston, not OpenAI marketing.
> How does Heisenberg's famous principle complicate this
It doesn't. Quantum uncertainty at subatomic scales has no demonstrated relevance to cognitive architecture. This is vibes.
> Companies... are claiming these tools do more than they are actually capable of
Possibly true! But "is cognition fundamentally predictive" is a question about brains, not LLMs. You've accidentally dismissed mainstream neuroscience while trying to critique AI hype.
> can you confidently classify all mental activity that you experience as "predicting the next thing"? [...] On what do you base your confidence in their equivalence?
To my understanding, bloaf's claim was only that the ability to predict seems a requirement of acting intentionally and thus that LLMs may "end up being a component in a system which actually does think" - not necessarily that all thought is prediction or that an LLM would be the entire system.
I'd personally go further and claim that correctly generating the next token is already a sufficiently general task to embed pretty much any intellectual capability. To complete `2360 + 8352 * 4 = ` for unseen problems is to be capable of arithmetic, for instance.
> When you have a thought, are you "predicting the next thing"—can you confidently classify all mental activity that you experience as "predicting the next thing"?
So notice that my original claim was "prediction is fundamental to our ability to act with intent" and now your demand is to prove that "prediction is fundamental to all mental activity."
That's a subtle but dishonest rhetorical shift to make me have to defend a much broader claim, which I have no desire to do.
> Language and society constrains the way we use words, but when you speak, are you "predicting"?
Yes, and necessarily so. One of the main objections that dualists use to argue that our mental processes must be immaterial is this [0]:
* If our mental processes are physical, then there cannot be an ultimate metaphysical truth-of-the-matter about the meaning of those processes.
* If there is no ultimate metaphysical truth-of-the-matter about what those processes mean, then everything they do and produce are similarly devoid of meaning.
* Asserting a non-dualist mind therefore implies your words are meaningless, a self-defeating assertion.
The simple answer to this dualist argument is precisely captured by this concept of prediction. There is no need to assert some kind of underlying magical meaning to be able to communicate. Instead, we need only say that in the relevant circumstances, our minds are capable of predicting what impact words will have on the receiver and choosing them accordingly. Since we humans don't have access to each other's minds, we must not learn these impacts from some kind of psychic mind-to-mind sense, but simply from observing the impacts of the words we choose on other parties; something that LLMs are currently (at least somewhat) capable of observing.
Exactly. Our base learning is by example, which is very much learning to predict.
Predict the right words, predict the answer, predict when the ball bounces, etc. Then reversing predictions that we have learned. I.e. choosing the action with the highest prediction of the outcome we want. Whether that is one step, or a series of predicted best steps.
Also, people confuse different levels of algorithm.
There are at least 4 levels of algorithm:
• 1 - The architecture.
This input-output calculation for pre-trained models are very well understood. We put together a model consisting of matrix/tensor operations and few other simple functions, and that is the model. Just a normal but high parameter calculation.
• 2 - The training algorithm.
These are completely understood.
There are certainly lots of questions about what is most efficient, alternatives, etc. But training algorithms harnessing gradients and similar feedback are very clearly defined.
• 3 - The type of problem a model is trained on.
Many basic problem forms are well understood. For instance, for prediction we have an ordered series of information, with later information to be predicted from earlier information. It could simply be an input and response that is learned. Or a long series of information.
• 4 - The solution learned to solve (3) the outer problem, using (2) the training algorithm on (1) the model architecture.
People keep confusing (4) with (1), (2) or (3). But it is very different.
For starters, in the general case, and for most any challenging problem, we never understand their solution. Someday it might be routine, but today we don't even know how to approach that for any significant problem.
Secondly, even with (1), (2), and (3) exactly the same, (4) is going to be wildly different based on the data characterizing the specific problem to solve. For complex problems, like language, layers and layers of sub-solutions to sub-problems have to be solved, and since models are not infinite in size, ways to repurpose sub-solutions, and weave together sub-solutions to address all the ways different sub-problems do and don't share commonalities.
Yes, prediction is the outer form of their solution. But to do that they have to learn all the relationships in the data. And there is no limit to how complex relationships in data can be. So there is no limit on the depths or complexity of the solutions found by successfully trained models.
Any argument they don't reason, based on the fact that they are being trained to predict, confuses at least (3) and (4). That is a category error.
It is true, they reason a lot more like our "fast thinking", intuitive responses, than our careful deep and reflective reasoning. And they are missing important functions, like a sense of what they know or don't. They don't continuously learn while inferencing. Or experience meta-learning, where they improve on their own reasoning abilities with reflection, like we do. And notoriously, by design, they don't "see" the letters that spell words in any normal sense. They see tokens.
Those reasoning limitations can be irritating or humorous. Like when a model seems to clearly recognize a failure you point out, but then replicates the same error over and over. No ability to learn on the spot. But they do reason.
Today, despite many successful models, nobody understands how models are able to reason like they do. There is shallow analysis. The weights are there to experiment with. But nobody can walk away from the model and training process, and build a language model directly themselves. We have no idea how to independently replicate what they have learned, despite having their solution right in front of us. Other than going through the whole process of retraining another one.
Every day I see people treat gen AI like a thinking human, Dijkstra's attitudes about anthropomorphizing computers is vindicated even more.
That said, I think the author's use of "bag of words" here is a mistake. Not only does it have a real meaning in a similar area as LLMs, but I don't think the metaphor explains anything. Gen AI tricks laypeople into treating its token inferences as "thinking" because it is trained to replicate the semiotic appearance of doing so. A "bag of words" doesn't sufficiently explain this behavior.
Yea bag of words isn’t helpful at all. I really do think that “superpowered sentence completion” is the best description. Not only is it reasonably accurate it is understandable, everyone has seen autocomplete function, and it’s useful. I don’t know how to “use” a bag of words. I do know how to use sentence completion. It also helps explains why context matters.
The contra-positive of "All LLMs are not thinking like humans" is "No humans are thinking like LLMs"
And I do not believe we actually understand human thinking well enough to make that assertion.
Indeed, it is my deep suspicion that we will eventually achieve AGI not by totally abandoning today's LLMs for some other paradigm, but rather embedding them in a loop with the right persistence mechanisms.
The loop, or more precisely the "search" does the novel part in thinking, the brain is just optimizing this process. Evolution could manage with the simplest model - copying with occasional errors, and in one run it made everyone of us. The moral - if you scale search the model can be dumb.
Bag of words is actually the perfect metaphor. The data structure is a bag. The output is a word. The selection strategy is opaquely undefined.
> Gen AI tricks laypeople into treating its token inferences as "thinking" because it is trained to replicate the semiotic appearance of doing so. A "bag of words" doesn't sufficiently explain this behavior.
Something about there being significant overlap between the smartest bears and the dumbest humans. Sorry you[0] were fooled by the magic bag.
[0] in the "not you, the layperson in question" sense
Spoken Query Language? Just like SQL, but for unstructured blobs of text as a database and unstructured language as a query? Also known as Slop Query Language or just Slop Machine for its unpredictable results.
> Spoken Query Language? Just like SQL, but for unstructured blobs of text as a database and unstructured language as a query?
I feel that's more a description of a search engine. Doesn't really give an intuition of why LLMs can do the things they do (beyond retrieval), or where/why they'll fail.
I am unsure myself whether we should regard LLMs as mere token-predicting automatons or as some new kind of incipient intelligence. Despite their origins as statistical parrots, the interpretability research from Anthropic [1] suggests that structures corresponding to meaning do exist inside those bundles of numbers and that there are signs of activity within those bundles of numbers that seem analogous to thought.
That said, I was struck by a recent interview with Anthropic’s Amanda Askell [2]. When she talks, she anthropomorphizes LLMs constantly. A few examples:
“I don't have all the answers of how should models feel about past model deprecation, about their own identity, but I do want to try and help models figure that out and then to at least know that we care about it and are thinking about it.”
“If you go into the depths of the model and you find some deep-seated insecurity, then that's really valuable.”
“... that could lead to models almost feeling afraid that they're gonna do the wrong thing or are very self-critical or feeling like humans are going to behave negatively towards them.”
Amanda Askell studied under David Chalmers at NYU: the philosopher who coined "the hard problem of consciousness" and is famous for taking phenomenal experience seriously rather than explaining it away. That context makes her choice to speak this way more striking: this isn't naive anthropomorphizing from someone unfamiliar with the debates. It's someone trained by one of the most rigorous philosophers of consciousness, who knows all the arguments for dismissing mental states in non-biological systems, and is still choosing to speak carefully about models potentially having something like feelings or insecurities.
> the interpretability research from Anthropic [1] suggests that structures corresponding to meaning do exist inside those bundles of numbers and that there are signs of activity within those bundles of numbers that seem analogous to thought
I did a simple experiment - took a photo of my kid in the park, showed it to Gemini and asked for a "detailed description". Then I took that description and put it into a generative model (Z-Image-Turbo, a new one). The output image was almost identical.
So one model converted image to text, the other reversed the processs. The photo was completely new, personal, never put online. So it was not in any training set. How did these 2 models do it if not actually using language like a thinking agent?
My fridge happily reads inputs without consciousness, has goals and takes decisions without "thinking", and consistently takes action to achieve those goals. (And it's not even a smart fridge! It's the one with a copper coil or whatever.)
I guess the cybernetic language might be less triggering here (talking about systems and measurements and control) but it's basically the same underlying principles. One is just "human flavored" and I therefore more prone to invite unhelpful lines of thinking?
Except that the "fridge" in this case is specifically and explicitly designed to emulate human behavior so... you would indeed expect to find structures corresponding to the patterns it's been designed to simulate.
Wondering if it's internalized any other human-like tendencies — having been explicitly trained to simulate the mechanisms that produced all human text — doesn't seem too unreasonable to me.
>research from Anthropic [1] suggests that structures corresponding to meaning exist inside those bundles of numbers and that there are signs of activity within those bundles of numbers that seem analogous to thought.
Can you give some concrete examples? The link you provided is kind of opaque
>Amanda Askell [2]. When she talks, she anthropomorphizes LLMs constantly.
She is a philosopher by trade and she describes her job (model alignment) as literally to ensure models "have good character traits." I imagine that explains a lot
Excerpt: “We found that there’s a specific combination of neurons in Claude’s neural network that activates when it encounters a mention (or a picture) of this most famous San Francisco landmark.”
Excerpt: “Recent research on smaller models has shown hints of shared grammatical mechanisms across languages. We investigate this by asking Claude for the ‘opposite of small’ across different languages, and find that the same core features for the concepts of smallness and oppositeness activate, and trigger a concept of largeness, which gets translated out into the language of the question.”
Excerpt: “Our new research provides evidence for some degree of introspective awareness in our current Claude models, as well as a degree of control over their own internal states.”
the anthropomorphization (say that 3 times quickly) is kinda weird, but also makes for a much more pleasant conversation imo. it's kinda tedious being pedantic all the time.
It also leads to fundamentally wrong conclusions: a related issue I have with this is the use of anthropomorphic shorthand when discussing international politics. You've heard a phrase like "the US thinks...", "China wants...", "Europe believes..." so much you don't even notice it.
All useful shorthands, all which lead to people displaying fundamental misunderstandings of what they're talking about - i.e. expressing surprise that a nation of millions doesn't display consistency of behavior of human lifetime scales, even though fairly obviously the mechanisms of government are churning their make up constantly, and depending on context maybe entirely different people.
It seems obvious to me that entities have emergent needs and plans and so on, independent of any of the humans inside.
For example, if you've worked at a large company, one of the little tragedies is when someone everyone likes gets laid off. There were probably no people who actively wanted Bob to lose his job. Even the CEO/Board who pulled the trigger probably had nothing against Bob. Heck, they might be the next ones out the door. The company is faceless, yet it wanted Bob to go, because that apparently contributed to the company's objective function. Had the company consisted entirely of different people, plus Bob, Bob might have been laid off anyway.
There is a strong will to do ... things the emerges from large structures of people and technology. It's funny like that.
I use LLMs heavily for work, I have done so for about 6 months. I see almost zero "thought" going on and a LOT of pattern matching. You can use this knowledge to your advantage if you understand this. If you're relying on it to "think", disaster will ensue. At least that's been my experience.
I've completely given up on using LLMs for anything more than a typing assistant / translator and maybe an encyclopedia when I don't care about correctness.
As usual with these, it helps to try to keep the metaphor used for downplaying AI, but flip the script. Let's grant the author's perception that AI is a "bag of words", which is already damn good at producing the "right words" for any given situation, and only keeps getting better at it.
Sure, this is not the same as being a human. Does that really mean, as the author seems to believe without argument, that humans need not be afraid that it will usurp their role? In how many contexts is the utility of having a human, if you squint, not just that a human has so far been the best way to "produce the right words in any given situation", that is, to use the meat-bag only in its capacity as a word-bag? In how many more contexts would a really good magic bag of words be better than a human, if it existed, even if the current human is used somewhat differently? The author seems to rest assured that a human (long-distance?) lover will not be replaced by a "bag of words"; why, especially once the bag of words is also ducttaped to a bag of pictures and a bag of sounds?
I can just imagine someone - a horse breeder, or an anthropomorphised horse - dismissing all concerns on the eve of the automotive revolution, talking about how marketers and gullible marks are prone to hippomorphising anything that looks like it can be ridden and some more, and sprinkling some anecdotes about kids riding broomsticks, legends of pegasi and patterns of stars in the sky being interpreted as horses since ancient times.
So a human is just a really expensive, unreliable bag of words. And we get more expensive and more unreliable by the day!
There's a quote I love but have misplaced, from the 19th century I think. "Our bodies are just contraptions for carrying our heads around." Or in this instance... bag of words transport system ;)
So tell me, why do I still have a job and why am frequently successful in getting profitable / useful products into production if I’m “expensive and unreliable”?
I mean I use AI tools to help achieve the goal but I don’t see any signs of the things I’m building and doing being unreliable.
I see a lot of people in tech claiming to "understand" what an LLM "really is" unlike all the gullible non-technical people out there. And, as one of those technical people who works in the LLM industry, I feel like I need call B.S. on us.
A. We don't really understand what's going on in LLMs. Mechanical interpretability is like a nascent field and the best results have come on dramatically smaller models. Understanding the surface-level mechanic of an LLM (an autoregressive transformer) should perhaps instill more wonder than confidence.
B. The field is changing quickly and is not limited to the literal mechanic of an LLM. Tool calls, reasoning models, parallel compute, and agentic loops add all kinds of new emergent effects. There are teams of geniuses with billion-dollar research budgets hunting for the next big trick.
C. Even if we were limited to baseline LLMs, they had very surprising properties as they scaled up and the scaling isn't done yet. GPT5 was based on the GPT4 pretraining. We might start seeing (actual) next-level LLMs next year. Who actually knows how that might go? <<yes, yes, I know Orion didn't go so well. But that was far from the last word on the subject.>>
As a consequence of my profession, I understand how LLMs work under the hood.
I also know that we data and tech folks will probably never win the battle over anthropomorphization.
The average user of AI, nevermind folks who should know better, is so easily convinced that AI "knows," "thinks," "lies," "wants," "understands," etc. Add to this that all AI hosts push this perspective (and why not, it's the easiest white lie to get the user to act so that they get a lot of value), and there's really too much to fight against.
We're just gonna keep on running into this and it'll just be like when you take chemistry and physics and the teachers say, "it's not actually like this but we'll get to how some years down the line- just pretend this is true for the time being."
These discussions often end up resembling religious arguments. "We don't know how any of this works, but we can fathom an intelligent god doing it, therefore an intelligent god did it."
"We don't really know how human consciousness works, but the LLM resembles things we associate with thought, therefore it is thought."
I think most people would agree that the functioning of an LLM resembles human thought, but I think most people, even the ones who think that LLMs can think, would agree that LLMs don't think in the exact same way that a human brain does. At best, you can argue that whatever they are doing could be classified as "thought" because we barely have a good definition for the word in the first place.
I'm a neurologist, and as a consequence of my profession, I understand how humans work under the hood.
The average human is so easily convinced that humans "know", "think", "lie", "want", "understand", etc.
But really it's all just a probabilistic chain reaction of electrochemical and thermal interactions. There is literally nowhere in the brain's internals for anything like "knowing" or "thinking" or "lying" to happen!
This is a fundamentally interesting point. Taking your comment as HN would advise, I totally agree.
I think genAI freaks a lot of people out because it makes them doubt what they thought made them special.
And to your comment, humans have always used words they reserve for humanity that indicates we're special: that we think, feel, etc... That we're human. Maybe we're not so special. Maybe that's scary to a lot of people.
There are no properties of matter or energy that can have a sense of self or experience qualia. Yet we all do. Denying the hard problem of consciousness just slows down our progress in discovering what it is.
It doesn't strike you as a bit...illogical to state in your first sentence that you "understand how humans work under the hood" and then go on to say that humans don't actually "understand" anything? Clearly everything at its basis is a chemical reaction, but the right reactions chained together create understanding, knowing, etc. I do believe that the human brain can be modeled by machines, but I don't believe LLMs are anywhere close to being on the right track.
My second thought is that it's not the metaphor that is misleading. People have been told thousands of times that LLMs don't "think", don't "know", don't "feel", but are "just a very impressive autocomplete". If they still really want to completely ignore that, why would they suddenly change their mind with a new metaphor?
Humans are lazy. If it looks true enough and it cost less effort, humans will love it. "Are you sure the LLM did your job correctly?" is completely irrelevant: people couldn't care less if it's correct or not. As long as the employer believes that the employee is "doing their job", that's good enough. So the question is really: "do you think you'll get fired if you use this?". If the answer is "no, actually I may even look more productive to my employer", then why would people not use it?
The problem with these metaphors is that they don't really explain anything. LLMs can solve countless problems today that we would have previously said were impossible because there are not enough examples in the training data. (EG, novel IMO/ICPC problems.) One way that we move the goal posts is to increase the level of abstraction: IMO/ICPC problems are just math problems, right? There are tons of those in the data set!
But the truth is there has been a major semantic shift. Previously LLMs could only solve puzzles whose answers were literally in the training data. It could answer a math puzzle it had seen before, but if you rephrased it only slightly it could no longer answer.
But now, LLMs can solve puzzles where, like, it has seen a certain strategy before. The newest IMO and ICPC problems were only "in the training data" for a very, very abstract definition of training data.
The goal posts will likely have to shift again, because the next target is training LLMs to independently perform longer chunks of economically useful work, interfacing with all the same tools that white-collar employees do. It's all LLM slop til it isn't, same as the IMO or Putnam exam.
And then we'll have people saying that "white collar employment was all in the training data anyway, if you think about it," at which point the metaphor will have become officially useless.
Yes, there are really two parallel claims here, aren't there: LLMs are not people (true, maybe true forever), and LLMs are only good at things that are well-represented in text form already. (false in certain categories and probably expanding to more in the future.)
Tokens in form of neural impulses go in, tokens in the form of neural impulses go out.
We would like to believe that there is something profound happening inside and we call that consciousness. Unfortunately when reading about split-brain patient experiments or agenesis of the corpus callosum cases I feel like we are all deceived, every moment of every day. I came to realization that the confabulation that is observed is just a more pronounced effect of the normal.
Could an LLM trained on nothing and looped upon itself eventually develop language, more complex concepts, and everything else, based on nothing? If you loop LLMs on each other, training them so they "learn" over time, will they eventually form and develop new concepts, cultures, and languages organically over time? I don't have an answer to that question, but I strongly doubt it.
There's clearly more going on in the human mind than just token prediction.
If you come up with a genetic algorithm scaffolding to affect both the architecture and the training algorithm, and then you instantiate it in an artificial selection environment, and you also give it trillions generations to evolve evolvability just right (as life had for billions of years) then the answer is yes, I'm certain it will and probably much sooner than we did.
Also, I think there is a very high chance that given an existing LLM architecture there exists a set of weights that would manifest a true intelligence immediately upon instantiation (with anterograde amnesia). Finding this set of weights is the problem.
I'm certain it wouldn't, and you're certain it would, and we have the same amount of evidence (and probably roughly the same means for running such an expensive experiment). I think they're more likely to go slowly mad, degrading their reasoning to nothing useful rather than building something real, but that could be different if they weren't detached from sensory input. Human minds looping for generations without senses, a world, or bodies might also go the same way.
> Also, I think there is a very high chance that given an existing LLM architecture there exists a set of weights that would manifest a true intelligence immediately upon instantiation (with anterograde amnesia).
I don't see why that would be the case at all, and I regularly use the latest and most expensive LLMs and am aware enough of how they work to implement them on the simplest level myself, so it's not just me being uninformed or ignorant.
The attention mechanism is capable of computing, in my thought experiment where you can magically pluck a weights-set from a trillion-dimensional space the tokens the machine will predict will only have a tiny subset dedicated to language. We have no capability of training such a system at this time, much like we have no way of training a non-differentiable architecture.
I would say that, token prediction is one of the things a brain does. And in a lot of people, most of what it does. But I dont think its the whole story. Possibly it is the whole story since the development of language.
"The machine accepts Chinese characters as input, carries out each instruction of the program step by step, and then produces Chinese characters as output. The machine does this so perfectly that no one can tell that they are communicating with a machine and not a hidden Chinese speaker.
The questions at issue are these: does the machine actually understand the conversation, or is it just simulating the ability to understand the conversation? Does the machine have a mind in exactly the same sense that people do, or is it just acting as if it had a mind?"
But even more than that, today’s AI chats are far more sophisticated than probabilistically producing the next word. Mixture of experts routes to different models. Agents are able to search the web, write and execute programs, or use other tools. This means they can actively seek out additional context to produce a better answer. They also have heuristics for deciding if an answer is correct or if they should use tools to try to find a better answer.
The article is correct that they aren’t humans and they have a lot of behaviors that are not like humans, but oversimplifying how they work is not helpful.
This is essentially Lady Lovelace's objection from the 19th century [1]. Turing addressed this directly in "Computing Machinery and Intelligence" (1950) [2], and implicitly via the halting problem in "On Computable Numbers" (1936) [3]. Later work on cellular automata, famously Conway's Game of Life [4], demonstrates more conclusively that this framing fails as a predictive model: simple rules produce structures no one "put in."
A test I did myself was to ask Claude (The LLM from Anthropic) to write working code for entirely novel instruction set architectures (e.g., custom ISAs from the game Turing Complete [5]), which is difficult to reconcile with pure retrieval.
I'm not convinced that "It's just a bag of words" would do much to sway someone who is overestimating an LLM's abilities. Feels too abstract/disconnected from what their experience using the LLM will be that it'll just sound obviously mistaken.
> But we don’t go to baseball games, spelling bees, and Taylor Swift concerts for the speed of the balls, the accuracy of the spelling, or the pureness of the pitch. We go because we care about humans doing those things.
My first thought was does anyone want to _watch_ me programming?
No, but watching a novelist at work is boring, and yet people like books that are written by humans because they speak to the condition of the human who wrote it.
Let us not forget the old saw from SICP, “Programs must be written for people to read, and only incidentally for machines to execute.” I feel a number of people in the industry today fail to live by that maxim.
It suggests to me, having encountered it for the first time, that programs must be readable to remain useful. Otherwise they'll be increasingly difficult to execute.
I vaguely remember a site where you could watch random people live streaming their programming environment, but I think twitch ate it, or maybe it was twitch -- not sure, but was interesting
I was trying to explain the concept of "token prediction" to my wife, whose eyes glaze over when discussing such technical topics. (I think she has the brainpower to understand them, but a horrible math teacher gave her a taste aversion to even attempting to that hasn't gone away. So she just buys Apple stuff and hopes Tim Apple hasn't shuffled around the UI bits AGAIN.)
I stumbled across a good-enough analogy based on something she loves: refrigerator magnet poetry, which if it's good consists of not just words but also word fragments like "s", "ed", and "ing" kinda like LLM tokens. I said that ChatGPT is like refrigerator magnet poetry in a magical bag of holding that somehow always gives the tile that's the most or nearly the most statistically plausible next token given the previous text. E.g., if the magnets already up read "easy come and easy ____", the bag would be likely to produce "go". That got into her head the idea that these things operate based on plausibility ratings from a statistical soup of words, not anything in the real world nor any internal cogitation about facts. Any knowledge or thought apparent in the LLM was conducted by the original human authors of the words in the soup.
Did you explain how LLMs can achieve gold-medal performance at math competitions involving original problems, without any original knowledge or thought?
Did she ask if a "statistical soup of words," if large enough, might somehow encode or represent something a little more profound than just a bunch of words?
The defenders and the critics around LLM anthropomorphism are both wrong.
The defenders are right insofar as the (very loose) anthropomorphizing language used around LLMs is justifiable to the extent that human beings also rely on disorder and stochastic processes for creativity. The critics are right insofar as equating these machines to humans is preposterous and mostly relies on significantly diminishing our notion of what "human" means.
Both sides fail to meet the reality that LLMs are their own thing, with their own peculiar behaviors and place in the world. They are not human and they are somewhat more than previous software and the way we engage with it.
However, the defenders are less defensible insofar as their take is mostly used to dissimulate in efforts to make the tech sound more impressive than it actually is. The critics at least have the interests of consumers and their full education in mind—their position is one that properly equips consumers to use these tools with an appropriate amount of caution and scrutiny. The defenders generally want to defend an overreaching use of metaphor to help drive sales.
Give it time. The first iPhone sucked compared to the Nokia/Blackberry flagships of the day. No 3G support, couldn't copy/paste, no apps, no GPS, crappy camera, quick price drops, negligible sales in the overall market.
Your analogy makes no sense. VHS spawned the entire home market, which went through multiple quality upgrades well above beta. It would only make sense if in 2025 we were using vhs everywhere and that the current state of the art for LLMs is all there ever is.
I feel like their analogy could have worked if they had pushed a little further into it.
The RNN and LSTM architectures (and Word2Vec, n-grams, etc) yielded language models that never got mass adoption. Like reel to reel. Then the transformer+attention hit the scene and several paths kicked off pretty close to each other. Google was working on Bert/encoder only transformer, maybe you could call that betamax. Doesn’t perfectly fit as in the case of beta it was actually the better tech.
OpenAI ran with the generative pre trained transformer and ML had its VHS? moment. Widespread adoption. Universal awareness within the populace.
Now with Titans (+miras?) are we entering the dvd era? Maybe. Learning context on the fly (memorizing at test time) is so much more efficient, it would be natural to call it a generational shift, but there is so much in the works right now with the promise of taking us further, this all might end up looking like the blip that beta vs vhs was. If current gen OpenAI type approaches somehow own the next 5-10 years then Titans, etc as Betamax starts to really fit - the shittier tech got and kept mass adoption. I don’t think that’s going to happen, but who knows.
Taking the analogy to present - who in the vhs or even earlier dvd days could imagine ubiquitous 4k+ vod? Who could have stood in a blockbuster in 2006 and knew that in less than 20 years all these stores and all these dvds would be a distant memory, completely usurped and transformed? Innovation of home video had a fraction of the capital being thrown at it that AI/ML has being thrown at it today. I would expect transformative generational shifts the likes of reel to cassette to optical to happen in fractions of the time they happened to home video. And beta/vhs type wars to begin and end in near realtime.
The mass adoption and societal transformation at the hands of AI/ML is just beginning. There is so. much. more. to. come. In 2030 we will look back at the state of AI in December 2025 and think “how quaint”, much the same as how we think of a circa 2006 busy Blockbuster.
Vhs came out in 76, blockbuster started in 85 (we went to video stores well before that when I was a kid), dvd in 95. I remember the sopranos making a joke about how dvd was barely taking off, they started in 99. Lets call it VHS had a run from 80 to 99, that's 19 years. The iphone launched in 2007, when did mobile become huge or inseprable from doing life (by force by so many apps), probbably in the pandemic.
I wouldn't say VHS was a blip. It was the recorded half video of media for almost 20 years.
I agree with the rest of what you said.
I'll say that the differences in the AI you're talking about today might be like the differences between VAX, PC JR, and the Lisa. All things before computing went main stream. I do think things go mainstream from tech a lot faster these days, people don't want to miss out.
I don't know where I'm going with this, I'm reading and replying to HN while watching the late night NFL game in an airport lounge.
Beta was superior in everything but run length, and it lost because it was more expensive than VHS without being sufficiently superior to justify the cost.
> That’s also why I see no point in using AI to, say, write an essay, just like I see no point in bringing a forklift to the gym. Sure, it can lift the weights, but I’m not trying to suspend a barbell above the floor for the hell of it. I lift it because I want to become the kind of person who can lift it. Similarly, I write because I want to become the kind of person who can think.
reply