[bertil]

> the AI says things like “Interesting!”

My experience of those utterance is that it’s purely phatic mimicry: they lack genuine intuitive surprise, it’s just marking a very odd shift in direction. The problem isn’t the lack of path, is that the rhetorical follow-up to those leaps are usually relevant results, so they stream-of-token ends up rapidly over-playing its own conviction. That’s why it’s necessary (and often ineffective) to tell them to validate their findings thoroughly: too much of their training is “That’s odd” followed by “Eureka!” and not “Nevermind…”

[sigbottle]

I think that a lot of models have to sprinkle in a lot of "fluff" in their thinking to stay within the right distribution. They only have language as their only medium; the way we annotate context is via brackets and then training them to hopefully respect the brackets. I'd imagine that either top labs explicitly train, or through the RL process the models implicitly learn, to spam tokens to keep them 'within distribution' since everything's going through the same channel and there's no fine grained separation between things.

Philosophically, it's not like you're a detached observer who simply reasons over all possible hypotheses. Ever get stuck in a dead end and find it hard to dig yourself out? If you were a detached observer, it'd be pretty easy to just switch gears. But it's not (for humans).

[WarmWash]

Language really only exists at the input and output surfaces of the models. In the middle it's all numerical values. Which you might be quick in relating to just being a numeric cypher of the words, which while not totally false, it misses that it is also a numeric cypher of anything. You can train a transformer on anything that you can assign tokens to.

[sigbottle]

That's not my point. I'm talking about something far more mundane - transformers do inference over raw tokens and perform an n^2 loop over tokens, but tokens are itself the context. So it's better to have more raw tokens in your input that all nudge it to the right idea space, even if technically it doesn't need all those tokens. ICL and CoT have a lot of study into them at this point, these are well known phenomena.

This applies to any transformer-based architecture including JEPA which tries to make the tokens predict some kind of latent space (in which I've separately heard arguments as to why the two are equivalent, but that's a different discussion.)

[pohl]

Similarly, none of our comments actually exist as language on Hacker News—just numerical values from the ASCII table. We're deluding each other into thinking we're using language.

[jfengel]

I believe it's reasonably clear that our thought processes generally occur outside of language. We do use language during explicit reasoning, but most thinking occurs heuristically. It's on par with the thinking of animals that don't use language but do complex behavior.

It not clear to me how well that maps onto LLMs. Our wetware predates language, and isn't derived from it. Language is built on top. LLMs are derived from language. I think that means that the intermediate layers are very different from the brain neurons, but I don't know. It's eerie how well the former emulates the latter.

[tonyarkles]

There’s an interesting thing there that I believe varies person to person. My understanding is that some people do think in a more symbolic/heuristic way, some rely very heavily on their inner monologue to make sense of things (I am in the latter camp, and only have a single core language processor so pretty much cannot come up with coherent thoughts if I’m concentrating on what someone else is saying)

Even more interesting, and getting off on a bit of a tangent, there is also a mode that I use for revealing emotions that I don’t have words for (alexythmia): I open up a text editor, stare off into space, and let my fingers type without “observing” the stream of words coming out. I then go back and read what I “wrote” and often end up understanding how I’m feeling much better than I did. It’s weird.

Edit: also, playing with local models through e.g. llama-cpp in “thinking mode” is super fascinating for me. The “thought process” that comes out before the real answer often feels pretty familiar when I reflect on my own inner monologue, although sometimes it’s frustrating for me because I see where their “thinking” went off the rails and want to correct it.

[alxndr]

"The great enemy of communication, we find, is the illusion of it" —William H. Whyte

[etherealG]

And what I find fascinating is I see similar mimicking by my 5 year old. Perhaps we shouldn’t be so quick to call this a lack of being genuine. Sometimes emotions are learned in humans but we wouldn’t call them fake.

I don’t want to declare machines to have emotion outright, but to call mimicry evidence of falsehood is also itself false.

[nkrisc]

Mimicry is how kids learn the expected reactions to particular emotions. A kid mimicking your surprise doesn’t mean they are surprised (as surprise requires an existing expectation of an outcome they may not have the experience for), but when they do feel genuine surprise, they’ll know how to express it.

[orangebread]

How do we know that AI isn't feeling genuine surprise then?

[philipallstar]

Because it's a statistical process generating one part of a word at a time. It probably isn't even generating "surprise". It might be generating "sur", then "prise" then "!"

[yulker]

But what is surprise really? Something not following expectation. The distribution may statistically leverage surprise as a concept via how it has seen surprise as a concept e.g. "interesting!"

So it can be both true that it has nothing to do with the emotion of surprise, but appear as the emulation of that emotion since the training data matches the concept of surprise (mismatch between expectation and event).

[nkrisc]

It’s the emotional and physiological response to a prediction being wrong. At its most primal, it’s the fear and surge of adrenaline when a predator or threat steps out from where you thought there was no threat. That’s not something most people will literally experience these days but even comedic surprise stems from that shock of subversion of expectation.

LLMs do not feel. They can express feeling, just as you can, but it doesn’t stem from a true source of feeling or sensation.

Expressing fake feelings is trivial for humans to do, and apparently for an LLM as well. I’m sure many autistic people or even anyone who’s been given a gift they didn’t like can relate to expressing feelings that they don’t actually feel, because expressing a feeling externally is not at all the same as actually feeling it. Instead it’s how we show our internal state to others, when we want to or can’t help it.

It is a mistake to equate artificial intelligence with sentience and humanity for moral reasons, if nothing else.

[Gareth321]

We are also technically a statistical process generating one part of a word at a time when we speak. Our neurons form the same kind of vectorised connections LLMs do. We are the product of repeated experiences - the same way training works.

Our brains are more advanced, and we may not experience the world the same way, but I think we have clearly created rudimentary digital consciousness.

[lee_ars]

Because it has no mind, no cognition, and nothing to "feel" with. Don't mistake programmatic mimicry for intention. That's just your own linguistic-forward primate cognition being fooled by the linguistic signals the training set and prompt are making the AI emit.

[nkoren]

I could describe the electrical and chemical signals within your neurons and synapses as proof that you are merely a series of electrochemical reactions, and can only mimic genuine thought.

[lee_ars]

You could do that if you wanted to ignore reality and be reductive to score points in an argument by purposefully conflating mimicry with intention, yes.

[nkrisc]

That is, by definition, genuine thought.

[saidnooneever]

most emotions in humans are learnt in self exploration, this is more obvious in kids.

first there is only good and bad, then more nuanced emotions based on increased understanding of the context in which they arise

[jackcarter]

It’s funny that this is probably due to bias in the training texts, right? Humans are way more likely to publish their “Eureka!” moments than their screwups… if they did, maybe models would’ve exhibit this behavior.

Now that AI labs have all these “Nevermind” texts to train on, maybe it’s getting easier to correct? (Would require some postprocessing to classify the AI outputs as successful or not before training)

[embedding-shape]

I think it's more explicit than that, part of post-training to enforce the kind of behavior, I don't think it's emergent but rather researchers steering it to do that because they saw the CoT gets slightly better if the model tries to doubt itself or cheer itself on. Don't recall if there was a paper outlining this, tried finding where I got this from but searches/LLMing turns up nothing so far.

[Forgeties79]

My understanding is that it’s the result of these companies making sure to keep you engaged/happy less than the result of data these companies train with.

I don’t know if it’s true or not but it certainly tracks given LLMs are way more polite than the average post on the internet lol

[rocqua]

I believe there might be more to it. Wasn't a big part of thinking or reasoning taking the response, replacing the final period with "Wait!" and then continuing? Which suggests that such words actually are important to the internals.

[fnordpiglet]

I think sometimes though there harness LLMs providing guidance. For instance I’ve seen recently coding agents doing an analysis then mid response saying “no wait, that’s not right” and course correcting. This feels implausible as an auto regressive rhetorical tick. LLM harnesses are widely used in advanced agentic systems and I’m sure the Pro level reasoning models exploit them extensively. I’m not saying this is what happened here, but there is a chance it was something injected by the hardness into its thinking.

[epolanski]

Interestingly this is strikingly similar to how my mind would process something I find genuinely interesting.

[hmontazeri]

The new Opus 4.7 thinks quite often with: Hmmmm…

Haha anyone else seen this?

[holoduke]

Indeed. I think it's the client. Not the model

[animal531]

I've somehow managed to train mine out of trying to fluff me up the whole time, its become very factual.

Overall it saves me a lot of time reading when it's just focusing on the details.