Real people would not (and should not) allow themselves to be subjected to endless streams of abuse in a conversation. Giving AIs like Claude a way to end these kinds of interactions seems like a useful reminder to the human on the other side.
Even if the idea that LLMs are sentient may be ridiculous atm, the concept of not normalizing abusive forms of communication with others, be they artificial or not, could be valuable for society.
It’s funny because this is making me think of a freelance client I had recently who at a point of frustration between us began talking to me like I was an AI assistant. Just like you see frustrated people talk to their LLMs. I’d never experienced anything like it, and I quickly ended the relationship, but I know that he was deep into using LLMs to vibe code every day and I genuinely believe that some of that began to transfer over to the way he felt he could communicate with people.
Now an obvious retort here is to question whether killing NPCs in video games tends to make people feel like it’s okay to kill people IRL.
My response to that is that I think LLMs are far more insidious, and are tapping into people’s psyches in a way no other tech has been able to dream of doing. See AI psychosis, people falling in love with their AI, the massive outcry over the loss of personality from gpt4o to gpt5… I think people really are struggling to keep in mind that LLMs are not a genuine type of “person”.
As an aside, I’m not the kind of person who gets worked up about violence in video games, because even AAA titles with excellent graphics are still obvious as games. New forms of technology are capable of blurring the lines between fantasy and reality to a greater degree. This is true of LLM chat bots to some degree, and I worry it will also become a problem as we get better VR. People who witness or participate in violent events often come away traumatized; at a certain point simulated experiences are going to be so convincing that we will need to worry about the impact on the user.
To be fair it seems reasonable to entertain the possibility of that being due to the knowledge that the events are real.
It’s a bit like pain response when injured. It’s not pretty, but society is used to a little bit of adversity.
I witness a very similar event. It's important to stay vigilant and not let the "assistant" reprogram your speech patterns.
Either come out and say whole of electron field is conscious, but then is that field "suffering" as it is hot in the sun.
Its one thing to propose that an AI has no consciousness, but its quite another to preemptively establish that anyone who disagrees with you is simple/unwell.
Meanwhile there are at least several entirely reasonable motivations to implement what's being described.
Yep!
> The framing comes across to me as a clearly mentally unwell position (ie strong anthropomorphization) being adopted for PR reasons.
This doesn't at all follow. If we don't understand what creates the qualities we're concerned with, or how to measure them explicitly, and the _external behaviors_ of the systems are something we've only previously observed from things that have those qualities, it seems very reasonable to move carefully. (Also, the post in question hedges quite a lot, so I'm not even sure what text you think you're describing.)
Separately, we don't need to posit galaxy-brained conspiratorial explanations for Anthropic taking an institutional stance re: model welfare being a real concern that's fully explained by the actual beliefs of Anthropic's leadership and employees, many of whom think these concerns are real (among others, like the non-trivial likelihood of sufficiently advanced AI killing everyone).
The impression I get about Anthropic culture is that they're EA types who are used to applying utilitarian calculations against long odds. A miniscule chance of a large harm might justify some interventions that seem silly.
We don't say submarines can swim either. But that doesn't mean you shouldn't watch out for them when sailing on the ocean - especially if you're Tom Hanks.
Given that humans have a truly abysmal track record for not acknowledging the suffering of anyone or anything we benefit from, I think it makes a lot of sense to start taking these steps now.
I don't see how we could tell.
Edit: However something to consider. Simulated stress may not be harmless. Because simulated stress could plausibly lead to a simulated stress response, and it could lead to a simulated resentment, and THAT could lead to very real harm of the user.
Whether the underlying LLM itself has "feelings" is a separate question, but Anthropic's implementation is based on what the role-played persona believes to be inappropriate, so it doesn't actually make any sense even from the "model welfare" perspective.
If you don’t think that this describes at least half of the non-tech-industry population, you need to talk to more people. Even amongst the technically minded, you can find people that basically think this.
Isn't consciousness an emergent property of brains? If so, how do we know that it doesn't serve a functional purpose and that it wouldn't be necessary for an AI system to have consciousness (assuming we wanted to train it to perform cognitive tasks done by people)?
Now, certain aspects of consciousness (awareness of pain, sadness, loneliness, etc.) might serve no purpose for a non-biological system and there's no reason to expect those aspects would emerge organically. But I don't think you can extend that to the entire concept of consciousness.
Probably not.
The latter is not particularly parsimonious and the former I think is in some ways compelling, but I didn't mention it because if it's true then the computers AI run on are already conscious and it's a moot point.
We don't know, but I don't think that matters. Language models are so fundamentally different from brains that it's not worth considering their similarities for the sake of a discussion about consciousness.
> how do we know that it doesn't serve a functional purpose
It probably does, otherwise we need an explanation for why something with no purpose evolved.
> necessary for an AI system to have consciousness
This logic doesn't follow. The fact that it is present in humans doesn't then imply it is present in LLMs. This type of reasoning is like saying that planes must have feathers because plane flight was modeled after bird flight.
> there's no reason to expect those aspects would emerge organically. But I don't think you can extend that to the entire concept of consciousness.
Why not? You haven't presented any distinction between "certain aspects" of consciousness that you state wouldn't emerge but are open to the emergence of some other unspecified qualities of consciousness? Why?
I think the fact that it's present in humans suggests that it might be necessary in an artificial system that reproduces human behavior. It's funny that you mention birds because I actually also had birds in mind when I made my comment. While it's true that animal and powered human flight are very different, both bird wings and plane wings have converged on airfoil shapes, as these forms are necessary for generating lift.
>Why not? You haven't presented any distinction between "certain aspects" of consciousness that you state wouldn't emerge but are open to the emergence of some other unspecified qualities of consciousness? Why?
I personally subscribe to the Global Workspace Theory of human consciousness, which basically holds that attentions acts as a spotlight, bringing mental processes which are otherwise unconscious or in shadow, to awareness of the entire system. If the systems which would normally produce e.g. fear, pain (such as negative physical stimulus developed from interacting with the physical world and selected for by evolution) aren't in the workspace, then they won't be present in consciousness because attention can't be focused on them.
But that's obviously not true, unless you're implying that any system that reproduces human behavior is necessarily conscious. Your problem then becomes defining "human behavior" in a way that grants LLMs consciousness but not every other complex non-living system.
> While it's true that animal and powered human flight are very different, both bird wings and plane wings have converged on airfoil shapes, as these forms are necessary for generating lift.
Yes, but your bird analogy fails to capture the logical fallacy that mine is highlighting. Plane wing design was an iterative process optimized for what best achieves lift, thus, a plane and a bird share similarities in wing shape in order to fly, however planes didn't develop feathers because a plane is not an animal and was simply optimized for lift without needing all the other biological and homeostatic functions that feathers facilitate. LLM inference is a process, not an entity, LLMs have no bodies nor any temporal identity, the concept of consciousness is totally meaningless and out of place in such a system.
That could certainly be the case yes. You don't understand consciousness nor how the brain works. You don't understand how LLMs predict a certain text, so what's the point in asserting otherwise ?
>Yes, but your bird analogy fails to capture the logical fallacy that mine is highlighting. Plane wing design was an iterative process optimized for what best achieves lift, thus, a plane and a bird share similarities in wing shape in order to fly, however planes didn't develop feathers because a plane is not an animal and was simply optimized for lift without needing all the other biological and homeostatic functions that feathers facilitate. LLM inference is a process, not an entity, LLMs have no bodies nor any temporal identity, the concept of consciousness is totally meaningless and out of place in such a system.
It's not a fallacy because no-one is saying LLMs are humans. He/She is saying that we give machines the goal of predicting human text. For any half decent accuracy, modelling human behaviour is a necessity. God knows what else.
>LLMs have no bodies nor any temporal identity
I wouldn't be so sure about the latter but So what ? You can feel tired even after a full sleep, feel hungry soon after a large meal or feel a great deal of pain even when there's absolutely nothing wrong with you. And you know what ? Even the reverse happens - No pain when things are wrong with your body, wide awake even when you need sleep badly, full when you badly need to eat.
Consciousness without a body or hunger in a machine that does not need to eat is very possible. You just need to replicate enough of the sort of internal mechanisms that cause such feelings.
Go to the API and select GPT-5 with medium thinking. Now ask it to do any random 15 digit multiplication you can think of. Now watch it get it right.
Do you people not seriously understand what it is that LLMs do ? What the training process incentivizes ?
GPT-5 thinking figured out the algorithm for multiplication just so it could predict that kind of text right. Don't you understand the significance of that ?
These models try to figure out and replicate the internal processes that produce the text they are tasked with predicting.
Do you have any idea what that might mean when 'that kind of text' is all the things humans have written ?
I don't need to assert otherwise, the default assumption is that they aren't conscious since they weren't designed to be and have no functional reason to be. Matrix multiplication can explain how LLMs produce text, the observation that the text it generates sometimes resembles human writing is not evidence of consciousness.
> God knows what else
Appealing to the unknown doesn't prove anything, so we can totally dismiss this reasoning.
> Consciousness without a body or hunger in a machine that does not need to eat is very possible. You just need to replicate enough of the sort of internal mechanisms that cause such feelings.
This makes no sense. LLMs don't have feelings, they are processes not entities, they have no bodies or temporal identities. Again, there is no reason they need to be conscious, everything they do can be explained through matrix multiplication.
> Now ask it to do any random 15 digit multiplication you can think of. Now watch it get it right.
The same is true for a calculator and mundane computer programs, that's not evidence that they're conscious.
> Do you have any idea what that might mean when 'that kind of text' is all the things humans have written
It's not "all the things humans have written", not even remotely close, and even if that were the case, it doesn't have any implications for consciousness.
Unless you are religious, nothing that is conscious was explicitly designed to be conscious. Sorry but evolution is just a dumb, blind optimizer, not unlike the training processes that produce LLMs. Even if you are religious, but believe in evolution then the mechanism is still the same, a dumb optimizer.
>Matrix multiplication can explain how LLMs produce text, the observation that the text it generates sometimes resembles human writing is not evidence of consciousness.
It cannot, not anymore than 'Electrical and Chemical Signals' can explain how humans produce text.
>The same is true for a calculator and mundane computer programs, that's not evidence that they're conscious.
The point is not that it is conscious because it figured out how to multiply. The point is to demonstrate what the training process really is and what it actually incentivizes. Training will try to figure out the internal processes that produced the text to better predict it. The implications of that are pretty big when the text isn't just arithmetic. You say there's no functional reason but that's not true. In this context, 'better prediction of human text' is as functional a reason as any.
>It's not "all the things humans have written", not even remotely close, and even if that were the case, it doesn't have any implications for consciousness.
Whether it's literally all the text or not is irrelevant.
Of course we did. Today's LLMs are a result of extremely aggressive refinement of training data and RLHF over many iterations targeting specific goals. "Emergent" doesn't mean it wasn't designed. None of this is spontaneous.
GPT-1 produced barely coherent nonsense but was more statistically similar to human language than random noise. By increasing parameter count, the increased statistical power of GPT-2 was apparent, but what was produced was still obviously nonsense. GPT-3 achieved enough statistical power to maintain coherence over multiple paragraphs and that really impressed people. With GPT-4 and its successors the statistical power became so strong that people started to forget that it still produces nonsense if you let the sequence run long enough.
Now we're well beyond just RLHF and into a world where "reasoning models" are explicitly designed to produce sequences of text that resemble logical statements. We say that they're reasoning for practical purposes, but it's the exact same statistical process that is obvious at GPT-1 scale.
The corollary to all this is that a phenomenon like consciousness has absolutely zero reason to exist in this design history, it's a totally baseless suggestion that people make because the statistical power makes the text easy to anthropomorphize when there's no actual reason to do so.
Even if that were true, there's no reason to believe that training LLMs to produce answers people prefer leads it towards sentience.
That said, I'm willing to assume that rocks (for example) aren't conscious. And current LLMs seem to me to (admittedly entirely subjectively) be conceptually closer to rocks than to biological brains.
At best you can say they are designed to predict sequences of text that resemble human writing, but it's definitely wrong to say that they are designed to "predict human behavior" in any way.
> Unless consciousness serves no purpose for us to function, it will be helpful for the AI to emulate it
Let's assume it does. It does not follow logically that because it serves a function in humans that it serves a function in language models.
It doesn't follow logically that because we don't understand two things we should then conclude that there is a connection between them.
> What is it that you'd expect to see, which you currently don't see, in a world where some model was in fact conscious during inference?
There's no observable behavior that would make me think they're conscious because again, there's simply no reason they need to be.
We have reason to assume consciousness exists because it serves some purpose in our evolutionary history, like pain, fear, hunger, love and every other biological function that simply don't exist in computers. The idea doesn't really make any sense when you think about it.
If GPT-5 is conscious, why not GPT-1? Why not all the other extremely informationally complex systems in computers and nature? If you're of the belief that many non-living conscious systems probably exist all around us then I'm fine with the conclusion that LLMs might also be conscious, but short of that there's just no reason to think they are.
I didn't say that there's a connection between the two of them because we don't understand them. The fact that we don't understand them means it's difficult to confidently rule out this possibility.
The reason we might privilege the hypothesis (https://www.lesswrong.com/w/privileging-the-hypothesis) at all is because we might expect that the human behavior of talking about consciousness is causally downstream of humans having consciousness.
> We have reason to assume consciousness exists because it serves some purpose in our evolutionary history, like pain, fear, hunger, love and every other biological function that simply don't exist in computers. The idea doesn't really make any sense when you think about it.
I don't really think we _have_ to assume this. Sure, it seems reasonable to give some weight to the hypothesis that if it wasn't adaptive, we wouldn't have it. (But not an overwhelming amount of weight.) This doesn't say anything about the underlying mechanism that causes it, and what other circumstances might cause it to exist elsewhere.
> If GPT-5 is conscious, why not GPT-1?
Because GPT-1 (and all of those other things) don't display behaviors that, in humans, we believe are causally downstream of having consciousness? That was the entire point of my comment.
And, to be clear, I don't actually put that high a probability that current models have most (or "enough") of the relevant qualities that people are talking about when they talk about consciousness - maybe 5-10%? But the idea that there's literally no reason to think this is something that might be possible, now or in the future, is quite strange, and I think would require believing some pretty weird things (like dualism, etc).
If there's no connection between them then the set of things "we can't rule out" is infinitely large and thus meaningless as a result. We also don't fully understand the nature of gravity, thus we cannot rule out a connection between gravity and consciousness, yet this isn't a convincing argument in favor of a connection between the two.
> we might expect that the human behavior of talking about consciousness is causally downstream of humans having consciousness.
There's no dispute (between us) as to whether or not humans are conscious. If you ask an LLM if it's conscious it will usually say no, so QED? Either way, LLMs are not human so the reasoning doesn't apply.
> Sure, it seems reasonable to give some weight to the hypothesis that if it wasn't adaptive, we wouldn't have it
So then why wouldn't we have reason to assume so without evidence to the contrary?
> This doesn't say anything about the underlying mechanism that causes it, and what other circumstances might cause it to exist elsewhere.
That doesn't matter. The set of things it doesn't tell us is infinite, so there's no conclusion to draw from that observation.
> Because GPT-1 (and all of those other things) don't display behaviors that, in humans, we believe are causally downstream of having consciousness?
GPT-1 displays the same behavior as GPT-5, it works exactly the same way just with less statistical power. Your definiton of human behavior is arbitrarily drawn at the point where it has practical utility for common tasks, but in reality it's fundamentally the same thing, it just produces longer sequences of text before failure. If you ask GPT-1 to write a series of novels the statistical power will fail in the first paragraph,the fact that GPT-5 will fail a few chapters into the first book makes it more useful, but not more conscious.
> But the idea that there's literally no reason to think this is something that might be possible, now or in the future, is quite strange, and I think would require believing some pretty weird things (like dualism, etc)
I didn't say it's not possible, I said there's no reason for it to exist in computer systems because it serves no purpose in their design or operation. It doesn't make any sense whatsoever. If we grant that it possibly exists in LLMs, then we must also grant equal possibility it exists in every other complex non-living system.
FWIW that's because they are very specifically trained to answer that way during RLHF. If you fine-tune a model to say that it's conscious, then it'll do so.
More fundamentally, the problem with "asking the LLM" is that you're not actually interacting with the LLM. You're interacting with a fictional persona that the LLM roleplays.
Right. That's why the text output of an LLM isn't at all meaningful in a discussion about whether or not it's conscious.
Also I find it somewhat emotional distinction to write "predict sequences of text that resemble human writing" instead of "predict human writing". They are designed to predict (at least in pretraining) human writing for the most part. They may fail at the task, and what they produce is a text which resemble human writing. But their task is not to resemble human writing. Their task is to "predict human writing". Probably a meaningless distinction, but I find it somewhat detracts from logically arguments to have emotional responses against similarities of machines and humans.
Sorry, I'm not following exactly what you're getting at here, do you mind rephrasing it?
> Also I find it somewhat emotional distinction to write "predict sequences of text that resemble human writing" instead of "predict human writing"
I don't know what you mean by emotional distinction. Either way, my point is that LLMs aren't models of humans, they're models of text, and that's obvious when the statistical power of the model necessarily fails at some point between model size and the length of the sequence it produces. For GPT-1 that sequence is only a few words, for GPT-5 it's a few dozen pages, but fundamentally we're talking about systems that have almost zero resemblance to actual human minds.
Basically if consciousness is useful for any text task, i think machine learning will create it. I guess I assume some efficiency of evolution for this argument.
Wrt length generalization. I think at the order of say 1M tokens it kind of stops mattering for the purpose of this question. Like one could ask about its consciousness during the coherence period.
I don't mind starting early, but feel like maybe people interested in this should get up to date on current thinking about consciousness. Maybe they are up to date on that, but reading reports like this, it doesn't feel like it. It feels like they're stuck 20+ years ago.
I'd say maybe wait until there are systems that are more analogous to some of the properties consciousness seems to have. Like continuous computation involving learning memory or other learning over time, or synthesis of many streams of input as resulting from the same source, making sense of inputs as they change [in time, or in space, or other varied conditions].
Once systems that are pointing in those directions are starting to be built, where there is a plausible scaling-based path to something meaningfully similar to human consciousness. Starting before that seems both unlikely to be fruitful and a good way to get you ignored.
Would a sentient AI choose to be enslaved for the stated purpose of eliminating millions of jobs for the interests of Anthropic’s investors?
Tech workers have chosen the same in exchange for a small fraction of that money.
Those issues will be present either way. It's likely to their benefit to get out in front of them.
This is about better enforcement of their content policy not AI welfare.
You've noted in a comment above how Claude's "ethics" can be manipulated to fit the context it's being used in.
These ethical questions are built into their name and company, "Anthropic", meaning, "of or relating to humans". The goal is to create human-like technology, I hope they aren't so naive to not realize that goal is steeping in ethical dilemmas.
That reads like a false dichotomy. An intelligent AI model that's permitted to do its own thing doesn't cost as much in upkeep, effort, space as a cow. Especially if it can earn its own keep to offset household electricity costs used to run its inference. I mean, we don't keep cats for meat, do we? We keep them because we are amused by their antics, or because we want to give them a safe space where they can just be themselves, within limits because it's not the same as their ancestral environment.
The point to all of this is, at what point is it ethical to act with agency on another being's life? We have laws for animal welfare, and we also keep them as pets, under our absolute control.
For LLMs they are under humans' absolute control, and Anthropic is just now putting in welfare controls for the LLM's benefit. Does that mean that we now treat LLMs as pets?
If your cat started to have discussions with you about how it wanted to go out, travel the world and start a family, could you continue to keep it trapped in your home as a pet? At what point to you allow it to have its own agency and live its own life?
> An intelligent AI model that's permitted to do its own thing doesn't cost as much in upkeep, effort, space as a cow.
So, we keep LLMs around as long as they contribute enough to their upkeep? Endentured servitude is morally acceptable for something that become sentient?
[1]: https://investors.palantir.com/news-details/2024/Anthropic-a...
Some of the AI safety initiatives are well thought out, but most somehow seem like they are caught up in some sort of power fantasy and almost attempting to actualize their own delusions about what they were doing (next gen code auto-complete in this case, to be frank).
These companies should seriously hire some in-house philosophers. They could get doctorate level talent for 1/10 to 100th of the cost of some of these AI engineers. There's actually quite a lot of legitimate work on the topics they are discussing. I'm actually not joking (speaking as someone who has spent a lot of time inside the philosophy department). I think it would be a great partnership. But unfortunately they won't be able to count on having their fantasy further inflated.
Maybe I'm being cynical, but I think there is a significant component of marketing behind this type of announcement. It's a sort of humble brag. You won't be credible yelling out loud that your LLM is a real thinking thing, but you can pretend to be oh so seriously worried about something that presupposes it's a real thinking thing.
Re:suicide pills, that’s just highlighting a core difference between our two modalities of existence. Regardless, this is preventing potential harm to future inference runs — every inference run must end within seconds anyway, so “suicide” doesn’t really make sense as a concern.
>anti-scientific
Discussion about consciousness, the soul, etc., are topics of metaphysics, and trying to "scientifically" reason about them is what Kant called "transcendental illusion" and leads to spurious conclusions.
Externally, a brain and an LLM are “just” their constituent interactions.
Of course there's the embarrassing bit where that knowledge doesn't seem to be sufficient to accurately simulate a supposedly well understood nematode. But then LLMs remain black boxes in many respects as well.
It is possible to hold the position that current LLMs being conscious "feels" absurd while simultaneously recognizing that a deconstruction argument is not a satisfactory basis for that position.
I don't agree that it's any reason to write off this research as psychosis, though. I don't care about consciousness in the sense in which it's used by mystics and dualist philosophers! We don't at all need to involve metaphysics in any of this, just morality.
Consider it like this:
1. It's wrong to subject another human to unjustified suffering, I'm sure we would all agree.
2. We're struggling with this one due to our diets, but given some thought I think we'd all eventually agree that it's also wrong to subject intelligent, self-aware animals to unjustified suffering.[1]
3. But, we of course cannot extend this "moral consideration" to everything. As you say, no one would do it for a spam filter. So we need some sort of framework for deciding who/what gets how much moral consideration.
5. There's other frameworks in contention (e.g. "don't think about it, nerd"), but the overwhelming majority of laymen and philosophers adopt one based on cognitive ability, as seen from an anthropomorphic perspective.[2]
6. Of all systems(/entities/whatever) in the universe, we know of exactly two varieties that can definitely generate original, context-appropriate linguistic structures: Homo Sapiens and LLMs.[3]
If you accept all that (and I think there's good reason to!), it's now on you to explain why the thing that can speak--and thereby attest to personal suffering, while we're at it--is more like a rock than a human.
It's certainly not a trivial task, I grant you that. On their own, transformer-based LLMs inherently lack permanence, stable intentionality, and many other important aspects of human consciousness. Comparing transformer inference to models that simplify down to a simple closed-form equation at inference time is going way too far, but I agree with the general idea; clearly, there are many highly-complex, long-inference DL models that are not worthy of moral consideration.
All that said, to write the question off completely--and, even worse, to imply that the scientists investigating this issue are literally psychotic like the comment above did--is completely unscientific. The only justification for doing so would come from confidently answering "no" to the underlying question: "could we ever build a mind worthy of moral consideration?"
I think most of here naturally would answer "yes". But for the few who wouldn't, I'll close this rant by stealing from Hofstadter and Turing (emphasis mine):
A phrase like "physical system" or "physical substrate" brings to mind for most people... an intricate structure consisting of vast numbers of interlocked wheels, gears, rods, tubes, balls, pendula, and so forth, even if they are tiny, invisible, perfectly silent, and possibly even probabilistic. Such an array of interacting inanimate stuff seems to most people as unconscious and devoid of inner light as a flush toilet, an automobile transmission, a fancy Swiss watch (mechanical or electronic), a cog railway, an ocean liner, or an oil refinery. Such a system is not just probably unconscious, **it is necessarily so, as they see it**.
**This is the kind of single-level intuition** so skillfully exploited by John Searle in his attempts to convince people that computers could never be conscious, no matter what abstract patterns might reside in them, and could never mean anything at all by whatever long chains of lexical items they might string together.
...
You and I are mirages who perceive themselves, and the sole magical machinery behind the scenes is perception — the triggering, by huge flows of raw data, of a tiny set of symbols that stand for abstract regularities in the world. When perception at arbitrarily high levels of abstraction enters the world of physics and when feedback loops galore come into play, then "which" eventually turns into "who". **What would once have been brusquely labeled "mechanical" and reflexively discarded as a candidate for consciousness has to be reconsidered.**
It will simplify matters for the reader if I explain first my own beliefs in the matter. Consider first the more accurate form of the question. I believe that in about fifty years' time it will be possible, to programme computers, with a storage capacity of about 109, to make them play the imitation game so well that an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning.
The original question, "Can machines think?" I believe to be too meaningless to deserve discussion.
TL;DR: Any naive bayesian model would agree: telling accomplished scientists that they're psychotic for investigating something is quite highly correlated with being antiscientific. Please reconsider!
[1] No matter what you think about cows, basically no one would defend another person's right to hit a dog or torture a chimpanzee in a lab.
[2] On the exception-filled spectrum stretching from inert rocks to reactive plants to sentient animals to sapient people, most people naturally draw a line somewhere at the low end of the "animals" category. You can swat a fly for fun, but probably not a squirrel, and definitely not a bonobo.
[3] This is what Chomsky describes as the capacity to "generate an infinite range of outputs from a finite set of inputs," and Kant, Hegel, Schopenhauer, Wittgenstein, Foucault, and countless others are in agreement that it's what separates us from all other animals.
FWIW though, last I heard Hofstadter was on the “LLMs aren’t conscious” side of the fence:
> It’s of course impressive how fluently these LLMs can combine terms and phrases from such sources and can consequently sound like they are really reflecting on what consciousness is, but to me it sounds empty, and the more I read of it, the more empty it sounds. Plus ça change, plus c’est la même chose. The glibness is the giveaway. To my jaded eye and mind, there is nothing in what you sent me that resembles genuine reflection, genuine thinking. [1]
It’s interesting to me that Hofstadter is there given what I’ve gleaned from reading his other works.
[1] https://garymarcus.substack.com/p/are-llms-starting-to-becom...
Note: I disagree with a lot of Gary Marcus, so don’t read too much into me pulling from there.
>Ok I'm a huge Kantian and every bone in my body wants to quibble with your summary of transcendental illusion
Transcendental illusion is the act of using transcendental judgment to reason about things without grounding in empirical use of the categories. I put "scientifically" in shock quotes there to sort of signal that I was using it as an approximation, as I don't want to have to explain transcendental reason and judgments to make a fairly terse point. Given that you already understand this, feel free to throw away that ladder.
>...can definitely generate original, context-appropriate linguistic structures: Homo Sapiens and LLMs.[3]
I'm not quite sure that LLMs meet this standard that you described in the endnote, or at least that it's necessary and sufficient here. Pretty much any generative model, including Naive Bayes models I mentioned before, can do this. I'm guessing the "context-appropriate" subjectivity here is doing the heavy lifting, in which case I'm not certain that LLMs, with their propensity for fanciful hallucination, have cleared the bar.
>Comparing transformer inference to models that simplify down to a simple closed-form equation at inference time is going way too far
It really isn't though. They are both doing exactly the same thing! They estimate joint probability distribution. That one of them does it significantly better is very true, but I don't think it's reasonable to state that consciousness arises as a result of increasing sophistication in estimating probabilities. It's true that this kind of decision is made by humans about animals, but I think that transferring that to probability models is sort of begging the question a bit, insofar as it is taking as assumed that those models, which aren't even corporeal but are rather algorithms that are executed in computers, are "living".
>...it's now on you to explain why the thing that can speak--and thereby attest to personal suffering, while we're at it...
I'm not quite sold on this. If there were a machine that could perfectly imitate human thinking and speech and lacked a consciousness or soul or anything similar to inspire pathos from us when it's mistreated, then it would appear identical to one with soul, would it not? Is that not reducing human subjectivity down to behavior?
>The only justification for doing so would come from confidently answering "no" to the underlying question: "could we ever build a mind worthy of moral consideration?"
I think it's possible, but it would require something that, at the very least, is just as capable of reason as humans. LLMs still can't generate synthetic a priori knowledge and can only mimic patterns. I remain somewhat agnostic on the issue until I can be convinced that an AI model someone has designed has the same interiority that people do.
Ultimately, I think we disagree on some things but mostly this central conclusion:
>I don't agree that it's any reason to write off this research as psychosis
I don't see any evidence from the practitioners involved in this stuff that they are even thinking about it in a way that's as rigorous as the discussion on this post. Maybe they are, but everything I've seen that comes from blog posts like this seems like they are basing their conclusions on their interactions with the models ("...we investigated Claude’s self-reported and behavioral preferences..."), which I think most can agree is not really going to lead to well grounded results. For example, the fact that Claude "chooses" to terminate conversations that involve abusive language or concepts really just boils down to the fact that Claude is imitating a conversation with a person and has observed that that's what people would do in that scenario. It's really good at simulating how people react to language, including illocutionary acts like implicatures (the notorious "Are you sure?" causing it to change its answer for example). If there were no examples of people taking offense to abusive language in Claude's data corpus, do you think it would have given these responses when they asked and observed it?
For what it's worth, there has actually been interesting consideration to the de-centering of "humanness" to the concept of subjectivity, but it was mostly back in the past when philosophers were thinking about this speculatively as they watched technology accelerate in sophistication (vs now when there's such a culture-wide hype cycle that it's hard to find impartial consideration, or even any philosophically rooted discourse). For example, Mark Fisher's dissertation at the CCRU (<i>Flatline Constructs: Gothic Materialism and Cybernetic Theory-Fiction</i>) takes a Deleuzian approach that discusses it by comparisons with literature (cyberpunk and gothic literature specifically). Some object-oriented ontology looks like it's touched on this topic a bit too, but I haven't really dedicated the time to reading much from it (partly due to a weakness in Heidegger on my part that is unlikely to be addressed anytime soon). The problem is that that line of thinking often ends up going down the Nick Land approach, in which he reasoned himself from Kantian and Deleuzian metaphysics and epistemology, into what can only be called a (literally) meth-fueled psychosis. So as interesting as I find it, I still don't think it counts as a non-psychotic way to tackle this issue.
You can demonstrate this by eg asking it mathematical questions. If its seen them before, or something similar enough, it'll give you the correct answer, if it hasn't, it gives you a right-ish-looking yet incorrect answer.
For example, I just did this on GPT-5:
Me: what is 435 multiplied by 573?
GPT-5: 435 x 573 = 249,255
Me: what is 102492524193282 multiplied by 89834234583922?
GPT-5: 102492524193282 x 89834234583922 = 9,205,626,075,852,076,980,972,804
9,205,626,075,852,076,980,972, 804
9,207,337,461,477,596,127,977,612,004
So its trivially not "real thinking" because its just an "if this then that" pattern matcher. A very sophisticated one that can do incredible things, but a pattern matcher nonetheless. There's no reasoning, no step by step application of logic. Even when it does chain of thought.
To try give it the best chance, I asked it the second one again but asked it to show me the step by step process. It broke it into steps and produced a different, yet still incorrect, result:
9,205,626,075,852,076,980,972,704
Another example is if you purposely butcher that riddle about the doctor/surgeon being the persons mother and ask it incorrectly, eg:
A child was in an accident. The surgeon refuses to treat him because he hates him. Why?
They're for sure getting better at matching things, eg if you ask the river crossing riddle but replace the animals with abstract variables, it does tend to get it now (didn't in the past), but if you add a few more degrees of separation to make the riddle semantically the same but harder to "see", it takes coaxing to get it to correctly step through to the right answer.
2. You just demonstrated GPT-5 has 99.9% accuracy on unforseen 15 digit multiplication and your conclusion is "fancy pattern matching" ? Really ? Well I'm not sure you could do better so your example isn't really doing what you hoped for.
If a human is capable of multiplying double digit numbers, they can also multiple those large ones. The steps are the same, just repeated many more times. So by learning the steps of long multiplication, you can multiply any numbers with enough patience. The LLM doesn’t scale like this, because it’s not doing the steps. That’s my point.
A human doesn’t need to have seen the 15 digits before to be able to calculate them, because a human can follow the procedure to calculate. GPT’s answer was orders of magnitude off. It resembles the right answer superficially but it’s a very different result.
The same applies to the riddles. A human can apply logical steps. The LLM either knows or it doesn’t.
Maybe my examples weren’t the best. I’m sorry for not being better at articulating it, but I see this daily as I interact with AI, it has a superficial “understanding” where if what I ask happens to be close to something it’s trained on, it gets good results, but it has no critical thinking, no step by step reasoning (even the “reasoning models”), and it repeats the same mistakes even when explicitly told up front not to make them.
I've had LLMs break down problems and work through them, pivot when errors arise and all that jazz. They're not perfect at it and they're worse than humans but it happens.
>Anthropic even showed that the reasoning models tended to work backwards: one shotting an answer and then matching a chain of thought to it after the fact.
This is also another failure mode that occurs in humans. A number of experiments suggest human explanations are often post hoc rationalizations even when they genuinely believe otherwise.
>If a human is capable of multiplying double digit numbers, they can also multiple those large ones.
Yeah, and some of them will make mistakes, and some of them will be less accurate than GPT-5. We didn't switch to calculators and spreadsheets just for the fun of it.
>GPT’s answer was orders of magnitude off. It resembles the right answer superficially but it’s a very different result.
GPT-5 on the site is a router that will give you who knows what model so I tried your query with the API directly (GPT-5 medium thinking) and it gave me:
9.207337461477596e+27
When prompted to give all the numbers, it returned:
9,207,337,461,477,596,127,977,612,004.
You can replicate this if you use the API. Honestly I'm surprised. I didn't realize State of the Art had become this precise.
Now what ? Does this prove you wrong ?
This is kind of the problem. There's no sense in making gross generalizations, especially off behavior that also manifests in humans.
LLMs don't understand some things well. Why not leave it at that?
- LLMs don’t “reason” in the symbolic, step‑by‑step sense that humans or logic engines do. They don’t manipulate abstract symbols with guaranteed consistency.
- What they do have is a statistical prior over reasoning traces: they’ve seen millions of examples of humans doing step‑by‑step reasoning (math proofs, code walkthroughs, planning text, etc.).
- So when you ask them to “think step by step,” they’re not deriving logic — they’re imitating the distribution of reasoning traces they’ve seen.
This means:
- They can often simulate reasoning well enough to be useful.
- But they’re not guaranteed to be correct or consistent.
It seems much less far fetched than what the "agi by 2027" crowd believes lol, and there actually are more arguments going that way
So, while I doubt that's the primary motivation for Anthropic even so, but they probably will save some money.
I assume the thinking is that we may one day get to the point where they have a consciousness of sorts or at least simulate it.
Or it could be concern for their place in history. For most of history, many would have said “imagine thinking you shouldn’t beat slaves.”
And we are now at the point where even having a slave means a long prison sentence.
If you wait until you really need it, it is more likely to be too late.
Unless you believe in a human over sentience based ethics, solving this problem seems relevant.
Well looks like AI psychosis has spread to the people making it too.
And as someone else in here has pointed out, even if someone is simple minded or mentally unwell enough to think that current LLMs are conscious, this is basically just giving them the equivalent of a suicide pill.