It is only surprising to those who refuse to understand how LLMs work and continue to anthropomorphise them. There is no being “truthful” here, the model has no concept of right or wrong, true or false. It’s not “lying” to you, it’s spitting out text. It just so happens that sometimes that non-deterministic text aligns with reality, but you don’t really know when and neither does the model.
You see binary failures all the time when doing function calls or JSON outputs.
That is… “please call this function” … does not call function
“calling JSON endpoint”… does not emit JSON
so from the article the tool generates hallucinations that the tool has used external stuff: but that was entirely fictitious. it does not know that this tool usage was fictitious and then sticks by its guns.
The workaround is to have verification steps, throw away “bad” answers. Instead of expecting one true output, expect a stream of results which have a yield (agriculture) of a certain amount. say 95% work, 5% garbage. never consider the results truly accurate, just “accurate enough”. Verify always
If tou ask it to draw a schematic thigns somehow get even worse.
But what it is good at is proposing ideas. So if you want to do a thing that could be solved by using a Gilbert cell, the chances it might mention a Gilbert Cell are realistically there.
But I am already having students coming by with LLM slob circuits asking why the don't work..
- This appears to be a regression relative to the GPT-series models which is specific to the o-series models. GPT-series models do not fabricate answers as often, and when they do they rarely double-down in the way o3 does. This suggests there's something specific in the way the o-series models are being trained that produces this behavior. By default I would have expected a newer model to fabricate actions less often rather than more!
- We found instances where the chain-of-thought summary and output response contradict each other: in the reasoning summary, o3 states the truth that e.g. "I don't have a real laptop since I'm an AI ... I need to be clear that I'm just simulating this setup", but in the actual response, o3 does not acknowledge this at all and instead fabricates a specific laptop model (with e.g. a "14-inch chassis" and "32 GB unified memory"). This suggests that the model does have the capability of recognizing that the statement is not true, and still generates it anyway. (See https://x.com/TransluceAI/status/1912617944619839710 and https://chatgpt.com/share/6800134b-1758-8012-9d8f-63736268b0... for details.)
The model emits text. What it's emitted before is part of the input to the next text generation pass. Since the training data don't usually include much text saying one thing then afterwards saying "that was super stupid, actually it's this other way" the model also is unlikely to generate a new token saying the last one was irrational.
If you wanted to train a model to predict the next sentence would be a contradiction of the previous you could do that. "True" and "correct" and "recognize" are not in the picture.
This is overly simplistic and demonstratably false - there's plenty of scenarios where a model will sell something false on purpose (e.g. when joking) and will tell you it was false with high probability correctly whether it was false or not after that.
However you want to frame it - there's clearly a more accurate than chance evaluation of truthfulness.
The AI doesn't reason in any real way. It's calculating the probability of the next word appearing in the training set conditioned on the context that came before, and in cases where there are multiple likely candidates it's picking one at random.
To the extent you want to claim intelligence from these systems, it's actually present in the training data. The intelligence is not emergent, it's encoded by humans in the training data. The weaker that signal is to the noise of random internet garbage, the more likely the AI will be to pick a random choice that's not True.
The claude paper showed, that it has some internal model when answering in different languages.
The process of learning can have effects in it, which is more than statistics. IF the training itself optimizes itself by having a internal model representation, than its no longer just statistics.
It also sounds like that humans are the origin of intelligence, but if humans do the same thing as LLM, and the only difference is, that we do not train LLMs from scratch (letting them discover the world, letting them inventing languages etc. but priming them with our world), than our intelligence was emergent and the LLMs one by proxy.
I don't need to know how it works internally, why it works internally.
What you (and parent post) are suggesting is that it is not intelligent based on its working. This is not a scientific take on the subject.
This is in fact how it works for medicine. A drug works because it has been shown to work based on statistical evidence. Even if we don't know how it works internally.
My problem with this attitude is that it's surprisingly accurate for humans, especially mentally disabled ones. While I agree that something is "missing" about how LLMs display their intelligence, I think it's wrong to say that LLMs are "just spitting out text, they're not intelligent". To me, it is very clear that LLM models do display intelligence, even if said intelligence is a bit deficient, and even if it weren't, it wouldn't be exactly the type of intelligence we see in people.
My point is, the phrase "AI" has been thrown around pointlessly for a while already. Marketing people would sell a simple 100-line programs with a few branches as "AI", but all common people would say that this intelligence is indeed just a gimmick. But when ChatGPT got released, something flipped. Something feels different about talking to ChatGPT. Most people see that there is some intelligence in there, and it's just a few old men yelling at the clouds "It's not intelligence! It's just statistical token generation!" as though these two were mutually exclusive.
Finally, I'd like to point out you're not "alive". You're just a very complex chemical reaction/physical interaction. Your entire life can be explained using organic chemistry and a bit of basic physics. Yet for some reason, most people decide not to think of life in this way. They attribute complex personalities and emotionaly to living beings, even though it's mostly hormones and basic chemistry again. Why?
Your internal experience of your life cannot be, though (this may change in the future).
"... o-series models are often prompted with previous messages without having access to the relevant reasoning. When asked questions that rely on their internal reasoning for previous steps, they must then come up with a plausible explanation for their behavior."
The fact is that humans do this all the time too -- their subconscious prompts them to do something, which they then do without reflecting or analyzing what their motivation might be. When challenged on it, they come up with a rationalization, not an actual reflected explanation.
The movie "Memento" is basically about how humans do this -- use faulty memories to rationalize stories for ourselves. At some point, a secondary character asks the main character, "And this fancy suit you're wearing, this car, where did they come from?" The main character (who is unable to form any long term memory) says, "I'm an insurance agent; my wife had insurance and I used the money from the payout to buy them." To which the secondary character says, "An in your grief, you went out and bought a Jaguar."
Not to give any spoilers, but that's not where the Jaguar came from, and the secondary character knows that.
"... finding a direction in activation space that satisfies logical consistency properties, such as that a statement and its negation have opposite truth values"
The LLM itself still has no idea of the truth or falsity of what it spits out. But you can more accurately retrieve yes/no answers to knowledge encoded in the model by using this specific trick - it's a validation step you can impose - making it less likely that the yes/no answer is wrong.
Really? LLMs are bullshit generators, but design. The surprising thing here is that people think that LLMs are "powerful at solving math tasks". (They're not.)
That's not really surprising either. We have evolved to recognize ourselves in our environment. We recognize faces and emotions in power outlets and lawn chairs. Recognizing intelligence in the outputs of LLMs is less surprising than that. But the fact that we recognize intelligence in LLMs implies intelligence in them just about as much as your power outlet is happy or sad because it looks that way to you.
Ask it to create a Typescript server side hello world.
It produces a JS example.
Telling it that's incorrect (but no more detail) results in it iterating all sorts of mistakes.
In 20 iterations it never once asked me what was incorrect.
In contrast, o4-mini asked me after 5, o4-mini-high asked me after 1, but narrowed the question to "is it incorrect due to choice of runtime?" rather than "what's incorrect?"
I told it to "ask the right question" based on my statement ("it is incorrect") and it correctly asked "what is wrong with it?" before I pointed out no Typescript types.
This is the critical thinking we need not just reasoning (incorrectly).
Well TS is a strict superset of JS so it’s technically correct (which is the best kind of correct) to produce JS when asked for a TS version. So you’re the one that’s wrong.
Try that one at your next standup and see how it goes over with the team
The fact is the training data has confused JS with TS so the LLM can't "get its head" around the semantic, not technical difference.
Also the secondary point wasn't just that it was "incorrect" it's the fact its reasoning was worthless unless it knew who to ask and the right questions to ask.
If somebody says to you something you know is right, is actually wrong, the first thing you ask them is "why do you think that?" not "maybe I should think of this from a new angle, without evidence of what is wrong".
It illustrates lack of critical thinking, and also shows you missed the point of the question. :D
However, OpenAI mentiones a Python tool multiple times in the system card [1], e.g.: "OpenAI o3 and OpenAI o4-mini combine state-of-the-art reasoning with full tool capabilities—web browsing, Python, [...]"
"The models use tools in their chains of thought to augment their capabilities; for example, cropping or transforming images, searching the web, or using Python to analyze data during their thought process."
I interpreted this to mean o3 does have access to a tool that enables it to run code. Is my understanding wrong?
(Interestingly, even in the ChatGPT UI the o3 model will sometimes state that it ran code on its personal MacBook Pro M2! https://x.com/TransluceAI/status/1912617941725847841)
Then if it says "I ran this code and it says X" we can easily verify. This is a big part of the reason I want LLMs to run code.
Weirdly I have seen Gemini write code and make claims about the output. I can see the code, the claims it makes about the output are correct. I do not think it could make these correct claims without running the code. But the UI doesn't show me this. To verify it, I have to run the code myself. This makes the whole feature way less valuable and I don't understand why!
I can vouch that this is extremely characteristic of o3-mini compared to competing models (Claude, Gemini) and previous OA models (3.5, 4o).
Compared to those, o3-mini clearly has less of the "the user is always right" training. This is almost certainly intentional. At times, this can be useful - it's more willing to call you out when you're wrong, and less likely to agree with something just because you suggested it. But this excessive stubbornness is the great downside, and it's been so prevalent that I stopped using o3-mini.
I haven't had enough time with o3 yet, but if it is indeed an evolution of o3-mini, it comes at no surprise it's very bad for this as well.
"List of mayors of my City X".
All OF THEM, get it wrong. Hallucinate the names, wrong dates, etc. The list is on wikipedia, and for sure they trained on that data, but they are not able to answer properly.
o3-mini? It just says it doesn't know lol
"In a few rigged demos, it even lies in more serious ways, like hiding evidence that it failed on a task, in order to get better ratings."
I don't want my LLM to excel in IMO or codeforces. I want it to understand my significantly easier but complex to state problem, think of solutions, understand its own issues and resolve it, rather than be passive agressive.
Well that and the whole field is filled with AI hypemen who "contribute" by asking ChatGPT about the quality and validity of some other GPT response.
(Not the mentioned LLMs here though.)
I do the rational acting, and it does the rest.
In the same way it helps a lot to try to understand what the correct model of an AI is in order that we can use it more productively. Certainly based on it's 'measurable properties' it does not behave like a reasonable human being. Some of the time it does, some of the time it goes completely off the rails. So there must be some other model that is more useful. "They are not rational actors, they can only generate plausible-looking texts." - seems to be more useful to me. "They are rational actors" - would be more like magical thinking which is not what got us to where we are today.
It claims to have run code on a Macbook because that's a plausible response from a human in this situation. It's basically trying to beat the Turing Test, but if you know it's a computer, it's obvious it's lying to you.
I'm not an expert, but it is a concept in these systems. Check out some videos on Deepseek's R1 paper. In particular there's a lot they did to incentivize the chain-of-thought reasoning process towards correct answers in "coding, mathematics, science, and logic reasoning" during reinforcement learning. I presume basically all the state of the art CoT reasoning models have some similar "correct and useful reasoning" portion in their RL tuning. This explains why models are getting better at math and code, but not as much at creative writing. As I understand it, everybody is pretty data limited, but it's much easier to generate synthetic training data where there is a right answer than it is to make good synthetic creative writing. It's also much easier to check that the model is answering those problems correctly during training, rather than waiting for human feedback via RLHF.
It seems that OpenAI forgot to make sure their critic model punished o3 for being wrong it claimed it had a laptop, lol.
Nothing is "recognized", nor is anything "an error". Nothing is "thinking" any more than it would be if the LLM just printed whether the next letter were more likely to be a vowel or consonant. Just because it's doing a better job modeling text doesn't magically make it be doing something that's not a text prediction function.
The LLM merchants are driving it though, by using pre-existing words for things that are not what they are saying they are.
It's amazing what they can do, but an LLM cannot know if what it outputs is true or correct, just statistically likely.
Yeah, if you’re into playing stupid mind games while not even being right.
If you stick to just voicing your needs, it’s fine. And I don’t think the TS/JS story shows a lack of reasoning that would be relevant for other use cases.
If I ask questions outside of the things I already know about (probably pretty common, right?), it's not playing mind games. It's only a 'gotcha' question with the added context, otherwise it's just someone asking a question and getting back a Monkey's Paw answer: "aha! See, it's technically a subset of TS.."
You might as well give it equal credit for code that doesn't compile correctly, since the author didn't explicitly ask.
Yes yes, language modelling ends up being surprisingly powerful at scale, but that doesn't make it not language modelling.
The surprise of the LLM AI was that they were somewhat truthful at all.
The big difference between us and LLMs, however, is that we grow up in the real world, where some things really are true, and others really are false, and where truths are really useful to convey information, and falsehoods usually aren't (except truths reported to others may be inconvenient and unwelcome, so we learn to recognize that and learn to lie). LLMs, however, know only text. Immense amounts of text, without any way to test or experience whether it's actually true or false, without any access to a real world to relate it to.
It's entirely possible that the only way to produce really human-level intelligent AI with a concept of truth, is to train them while having them grow up in the real world in a robot body over a period of 20 years. And that would really restrict the scalability of AI.
These kids couldn't understand that the plastic garbage in their own nature is not part of nature.
Nonetheless, depending on what rules you mean, there are a lot of people who show that logic or 'truth' is not the same for everyone.
People believing in a god, ghosts, conspiricy theories, flat earth etc.
I'm more curious if the 'self' can only be trained if you have a clear line of control. We learn what the self is because there is a part which we can control and than there is a part which we can't control.
Of course there have still been plenty of meaningful innovations, like the transformer/attention thing, but it's mostly the fact that affordable graphics cars offer massively-parallel floating point calculations which turns out to be exactly what we need to scale this up. That and the sheer amount of data that's become available in the age of the Internet.
It's certainly important but this reads as overly simplistic to me. All the hardware we have today won't make an SVM or a random forest scale the way transformers do.