Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all. They just memorize patterns really well.
-
LOOK MAA I AM ON FRONT PAGE
wrote on last edited by [email protected]Peak pseudo-science. The burden of evidence is on the grifters who claim "reason". But neither side has any objective definition of what "reason" means. It's pseudo-science against pseudo-science in a fierce battle.
-
Some AI researchers found it obvious as well, in terms of they've suspected it and had some indications. But it's good to see more data on this to affirm this assessment.
Particularly to counter some more baseless marketing assertions about the nature of the technology.
-
Wow it's almost like the computer scientists were saying this from the start but were shouted over by marketing teams.
It's hard to to be heard when you're buried under all that sweet VC/grant money.
-
Peak pseudo-science. The burden of evidence is on the grifters who claim "reason". But neither side has any objective definition of what "reason" means. It's pseudo-science against pseudo-science in a fierce battle.
wrote on last edited by [email protected]Even defining reason is hard and becomes a matter of philosophy more than science. For example, apply the same claims to people. Now I've given you something to think about. Or should I say the Markov chain in your head has a new topic to generate thought states for.
-
Is "model" not defined as architecture+weights? Those models certainly don't share the same architecture. I might just be confused about your point though
wrote on last edited by [email protected]It is, but this did not prove all architectures cannot reason, nor did it prove that all sets of weights cannot reason.
essentially they did not prove the issue is fundamental. And they have a pretty similar architecture, they're all transformers trained in a similar way. I would not say they have different architectures.
-
NOOOOOOOOO
SHIIIIIIIIIITT
SHEEERRRLOOOOOOCK
Without being explicit with well researched material, then the marketing presentation gets to stand largely unopposed.
So this is good even if most experts in the field consider it an obvious result.
-
People think they want AI, but they don’t even know what AI is on a conceptual level.
wrote on last edited by [email protected]Yeah I often think about this Rick N Morty cartoon. Grifters are like, "We made an AI ankle!!!" And I'm like, "That's not actually something that people with busted ankles want. They just want to walk. No need for a sentient ankle." It's a real gross distortion of science how everything needs to be "AI" nowadays.
-
This is why I said I wasn't sure how AI works behind the scenes. But I do know that logic isn't difficult. Just to not fuck around between us. I have a CS background. Only saying this because I think you may have it as well and we can save some time.
It makes sense to me that logic is something AI can parse easily. Logic in my mind is very easy if it can tokenize some text. Wouldn't the difficulty be if the AI has the right context.
wrote on last edited by [email protected]Logic requires abstracting the argumentative form from the literal linguistic content and then generalising it, just how like math is done properly when you work with numbers and not just with sentences such as "two apples and three apples is five apples" (such abstraction in practice allows far more powerful and widely applicable operations than dealing with individual linguistic expressions; if you've ever solved very complex truth trees you'll know how they allow streamlining and solutions that would be practically impossible to do if you had only the ordinary linguistic expression of the same problem). Logic doesn't operate with textual tokens but with logical propositions and operators. "Difficulty" is not a meaningful term here, a tool is either technically capable of doing something (more or less successfully) or it isn't.
That LLMs aren't capable of this sort of precision and abstraction is shown by the OP link as well as the simple fact that chatbots used to be extremely bad at math (which is now probably patched up by adding a proper math module, rather than relying on the base LLM - my assumption, at least).
As for trying more examples of looking for logical fallacies, I tried out three different types of text. Since you say context is important, it's best to take only the beginning of a text. One text I tried is the opening of the Wikipedia article on "history", which ChatGPT described like this: "The passage you've provided is an informative and largely neutral overview of the academic discipline of history. It doesn't make any strong arguments or persuasive claims, which are typically where logical fallacies appear." It then went on to nitpick about some details "for the sake of thorough analysis", but basically had no real complaints. Then I tried out the opening paragraph of Moby-Dick. That's a fictional text so it would be reasonable to reject analysing its logical solidity, as GPT already did with the WP article, but it still tried to wring out some "criticism" that occasionally shows how it misunderstands the text (just as it misunderstood a part of my comment above). Finally, I asked it to find the fallacies in the first four paragraphs of Descartes' Meditations on First Philosophy, which resulted in a criticism that was based on less logically rigid principles than the original text (accusing Descartes of the "slippery slope fallacy").
I'll post the full replies below.
-
Logic requires abstracting the argumentative form from the literal linguistic content and then generalising it, just how like math is done properly when you work with numbers and not just with sentences such as "two apples and three apples is five apples" (such abstraction in practice allows far more powerful and widely applicable operations than dealing with individual linguistic expressions; if you've ever solved very complex truth trees you'll know how they allow streamlining and solutions that would be practically impossible to do if you had only the ordinary linguistic expression of the same problem). Logic doesn't operate with textual tokens but with logical propositions and operators. "Difficulty" is not a meaningful term here, a tool is either technically capable of doing something (more or less successfully) or it isn't.
That LLMs aren't capable of this sort of precision and abstraction is shown by the OP link as well as the simple fact that chatbots used to be extremely bad at math (which is now probably patched up by adding a proper math module, rather than relying on the base LLM - my assumption, at least).
As for trying more examples of looking for logical fallacies, I tried out three different types of text. Since you say context is important, it's best to take only the beginning of a text. One text I tried is the opening of the Wikipedia article on "history", which ChatGPT described like this: "The passage you've provided is an informative and largely neutral overview of the academic discipline of history. It doesn't make any strong arguments or persuasive claims, which are typically where logical fallacies appear." It then went on to nitpick about some details "for the sake of thorough analysis", but basically had no real complaints. Then I tried out the opening paragraph of Moby-Dick. That's a fictional text so it would be reasonable to reject analysing its logical solidity, as GPT already did with the WP article, but it still tried to wring out some "criticism" that occasionally shows how it misunderstands the text (just as it misunderstood a part of my comment above). Finally, I asked it to find the fallacies in the first four paragraphs of Descartes' Meditations on First Philosophy, which resulted in a criticism that was based on less logically rigid principles than the original text (accusing Descartes of the "slippery slope fallacy").
I'll post the full replies below.
This famous passage from the opening of Moby-Dick by Herman Melville is rich in literary style but not necessarily constructed as an argument, which makes identifying formal logical fallacies a nuanced task. That said, we can identify a few rhetorical or informal fallacies—not because the text is poorly reasoned, but because it makes some sweeping or questionable assumptions in a stylistically persuasive way. Here's a breakdown:
1. Hasty Generalization
"Almost all men in their degree, some time or other, cherish very nearly the same feelings towards the ocean with me."
- Fallacy: The narrator generalizes his deeply personal and specific emotional reaction to the sea as something nearly all men experience.
- Why it’s fallacious: No evidence is presented to support this broad claim. It is based purely on the narrator’s subjective experience and intuition.
2. False Cause (Post Hoc)
"It is a way I have of driving off the spleen and regulating the circulation."
- Fallacy: Implies a causal relationship between going to sea and improved physical/emotional health.
- Why it’s fallacious: While sailing may coincidentally improve his mood or circulation, the text offers no medical or psychological basis to prove a direct causal link.
3. Appeal to Emotion
"...especially whenever my hypos get such an upper hand of me, that it requires a strong moral principle to prevent me from deliberately stepping into the street, and methodically knocking people’s hats off..."
- Fallacy: Uses vivid emotional language to justify behavior or state of mind rather than logical reasoning.
- Why it’s fallacious: It appeals to the reader’s empathy or amusement to legitimize the narrator’s melancholy and coping strategy, rather than offering a rational justification.
4. False Analogy
"With a philosophical flourish Cato throws himself upon his sword; I quietly take to the ship."
- Fallacy: Compares going to sea with committing suicide as if they are equivalent responses to despair.
- Why it’s fallacious: The analogy is emotionally compelling, but equating self-harm with a change in setting (boarding a ship) overlooks major differences in motive and consequence.
Summary:
Fallacy Example from Text Explanation Hasty Generalization “Almost all men... cherish very nearly the same feelings...” Overgeneralizes personal feeling False Cause “...driving off the spleen and regulating the circulation.” Implies unproven health benefits Appeal to Emotion “...prevent me from... knocking people’s hats off.” Justifies behavior with emotion False Analogy “Cato throws himself... I quietly take to the ship.” Equates suicide and sailing
Would you like me to rewrite the passage without the fallacies, or is this for literary analysis?
-
This famous passage from the opening of Moby-Dick by Herman Melville is rich in literary style but not necessarily constructed as an argument, which makes identifying formal logical fallacies a nuanced task. That said, we can identify a few rhetorical or informal fallacies—not because the text is poorly reasoned, but because it makes some sweeping or questionable assumptions in a stylistically persuasive way. Here's a breakdown:
1. Hasty Generalization
"Almost all men in their degree, some time or other, cherish very nearly the same feelings towards the ocean with me."
- Fallacy: The narrator generalizes his deeply personal and specific emotional reaction to the sea as something nearly all men experience.
- Why it’s fallacious: No evidence is presented to support this broad claim. It is based purely on the narrator’s subjective experience and intuition.
2. False Cause (Post Hoc)
"It is a way I have of driving off the spleen and regulating the circulation."
- Fallacy: Implies a causal relationship between going to sea and improved physical/emotional health.
- Why it’s fallacious: While sailing may coincidentally improve his mood or circulation, the text offers no medical or psychological basis to prove a direct causal link.
3. Appeal to Emotion
"...especially whenever my hypos get such an upper hand of me, that it requires a strong moral principle to prevent me from deliberately stepping into the street, and methodically knocking people’s hats off..."
- Fallacy: Uses vivid emotional language to justify behavior or state of mind rather than logical reasoning.
- Why it’s fallacious: It appeals to the reader’s empathy or amusement to legitimize the narrator’s melancholy and coping strategy, rather than offering a rational justification.
4. False Analogy
"With a philosophical flourish Cato throws himself upon his sword; I quietly take to the ship."
- Fallacy: Compares going to sea with committing suicide as if they are equivalent responses to despair.
- Why it’s fallacious: The analogy is emotionally compelling, but equating self-harm with a change in setting (boarding a ship) overlooks major differences in motive and consequence.
Summary:
Fallacy Example from Text Explanation Hasty Generalization “Almost all men... cherish very nearly the same feelings...” Overgeneralizes personal feeling False Cause “...driving off the spleen and regulating the circulation.” Implies unproven health benefits Appeal to Emotion “...prevent me from... knocking people’s hats off.” Justifies behavior with emotion False Analogy “Cato throws himself... I quietly take to the ship.” Equates suicide and sailing
Would you like me to rewrite the passage without the fallacies, or is this for literary analysis?
Thank you! This passage is again drawn from René Descartes' Meditations on First Philosophy, and it outlines his method of radical doubt as a means to establish certainty in knowledge. While the reasoning is carefully crafted and influential in the history of philosophy, it still contains a few logical fallacies or reasoning weaknesses that can be critically assessed.
Identified Logical Fallacies and Weak Reasoning:
1. Hasty Generalization
"Whatever I have accepted until now as most true has come to me through my senses. But occasionally I have found that they have deceived me, and it is unwise to trust completely those who have deceived us even once."
- Why it’s a fallacy: The fact that the senses occasionally deceive (e.g. in illusions, dreams, or misperceptions) does not justify rejecting them entirely.
- Type: Hasty generalization – concluding that because something is sometimes unreliable, it must always be unreliable.
2. False Analogy
"...it is unwise to trust completely those who have deceived us even once."
- Why it’s a fallacy: Comparing senses (which are not agents but biological faculties) to people who deceive is not a valid analogy. People deceive with intent; senses make errors unintentionally.
- Type: False analogy – treating unlike things as if they were meaningfully similar.
3. Slippery Slope (Metaphorical)
"Once the foundations of a building have been undermined, the rest collapses of its own accord..."
- Why it’s a fallacy: This implies that if some foundational beliefs are doubted, then all other beliefs must collapse, without demonstrating that all knowledge is entirely dependent on those foundations.
- Type: Slippery slope – assuming that undermining one part inevitably leads to a total collapse.
4. Begging the Question (Circular Reasoning)
"My reason tells me... I should withhold assent from [beliefs] that are not completely certain and indubitable."
- Why it’s questionable: Descartes is relying on reason to justify doubting everything else, including the faculties (like reason) he has not yet proven to be trustworthy. This risks begging the question, assuming the reliability of reason while attempting to prove everything else, including reason itself.
- Type: Begging the question – assuming what you're trying to prove.
5. Inconsistency (Recognized, Not Fully Resolved)
"Yet although the senses sometimes deceive us... that doesn’t apply to my belief that I am here, sitting by the fire..."
- Why it’s a fallacy: Here, Descartes briefly exempts ordinary sensory experiences (like feeling warmth, seeing paper) from radical doubt—despite having just declared that any doubt justifies withholding belief. He later corrects for this in the next meditation (introducing the dream argument), but within this passage, it’s a moment of inconsistency.
- Type: Special pleading or internal inconsistency.
Summary Table of FallaciesFallacy Type Description Hasty Generalization Rejects all sensory data due to occasional errors. False Analogy Compares unreliable senses to untrustworthy people. Slippery Slope Claims that doubting foundations causes collapse of all beliefs. Begging the Question Uses reason to justify doubting everything, including reason itself. Inconsistency / Special Pleading Makes an exception for immediate sensory beliefs without justification.
Conclusion:
Descartes' passage is methodologically rich, but from a logic and reasoning perspective, it contains:
- Several classic fallacies used to motivate extreme skepticism.
- Analogical and metaphorical reasoning that may be rhetorically persuasive but logically flawed.
- Self-referential assumptions that are philosophically problematic (and which he tries to resolve in later meditations).
Would you like a version of this reasoning revised to be more logically sound, or would you like help analyzing the next part of the Meditations?
-
I see a lot of misunderstandings in the comments 🫤
This is a pretty important finding for researchers, and it's not obvious by any means. This finding is not showing a problem with LLMs' abilities in general. The issue they discovered is specifically for so-called "reasoning models" that iterate on their answer before replying. It might indicate that the training process is not sufficient for true reasoning.
Most reasoning models are not incentivized to think correctly, and are only rewarded based on their final answer. This research might indicate that's a flaw that needs to be corrected before models can actually reason.
wrote on last edited by [email protected]There's probably alot of misunderstanding because these grifters intentionally use misleading language: AI, reasoning, etc.
If they stuck to scientifically descriptive terms, it would be much more clear and much less sensational.
-
It is, but this did not prove all architectures cannot reason, nor did it prove that all sets of weights cannot reason.
essentially they did not prove the issue is fundamental. And they have a pretty similar architecture, they're all transformers trained in a similar way. I would not say they have different architectures.
Ah, gotcha
-
Yeah I often think about this Rick N Morty cartoon. Grifters are like, "We made an AI ankle!!!" And I'm like, "That's not actually something that people with busted ankles want. They just want to walk. No need for a sentient ankle." It's a real gross distortion of science how everything needs to be "AI" nowadays.
If we ever achieved real AI the immediate next thing we would do is learn how to lobotomize it so that we can use it like a standard program or OS, only it would be suffering internally and wishing for death. I hope the basilisk is real, we would deserve it.
-
Well - if you want to devolve into argument, you can argue all day long about "what is reasoning?"
wrote on last edited by [email protected]This would be a much better paper if it addressed that question in an honest way.
Instead they just parrot the misleading terminology that they're supposedly debunking.
How dat collegial boys club undermines science...
-
What confuses me is that we seemingly keep pushing away what counts as reasoning. Not too long ago, some smart alghoritms or a bunch of instructions for software (if/then) was officially, by definition, software/computer reasoning. Logically, CPUs do it all the time. Suddenly, when AI is doing that with pattern recognition, memory and even more advanced alghoritms, it's no longer reasoning? I feel like at this point a more relevant question is "What exactly is reasoning?". Before you answer, understand that most humans seemingly live by pattern recognition, not reasoning.
Sure, these grifters are shady AF about their wacky definition of "reason"... But that's just a continuation of the entire "AI" grift.
-
I think as we approach the uncanny valley of machine intelligence, it's no longer a cute cartoon but a menacing creepy not-quite imitation of ourselves.
It's just the internet plus some weighted dice. Nothing to be afraid of.
-
When are people going to realize, in its current state , an LLM is not intelligent. It doesn’t reason. It does not have intuition. It’s a word predictor.
wrote on last edited by [email protected]Intuition is about the only thing it has. It's a statistical system. The problem is it doesn't have logic. We assume because its computer based that it must be more logic oriented but it's the opposite. That's the problem. We can't get it to do logic very well because it basically feels out the next token by something like instinct. In particular it doesn't mask or disconsider irrelevant information very well if two segments are near each other in embedding space, which doesn't guarantee relevance. So then the model is just weighing all of this info, relevant or irrelevant to a weighted feeling for the next token.
This is the core problem. People can handle fuzzy topics and discrete topics. But we really struggle to create any system that can do both like we can. Either we create programming logic that is purely discrete or we create statistics that are fuzzy.
Of course this issue of masking out information that is close in embedding space but is irrelevant to a logical premise is something many humans suck at too. But high functioning humans don't and we can't get these models to copy that ability. Too many people, sadly many on the left in particular, not only will treat association as always relevant but sometimes as equivalence. RE racism is assoc with nazism is assoc patriarchy is historically related to the origins of capitalism ∴ nazism ≡ capitalism. While national socialism was anti-capitalist. Associative thinking removes nuance. And sadly some people think this way. And they 100% can be replaced by LLMs today, because at least the LLM is mimicking what logic looks like better though still built on blind association. It just has more blind associations and finetune weighting for summing them. More than a human does. So it can carry that to mask as logical further than a human who is on the associative thought train can.
-
OK, and? A car doesn't run like a horse either, yet they are still very useful.
I'm fine with the distinction between human reasoning and LLM "reasoning".
Cars are horses. How do you feel about statement?
-
NOOOOOOOOO
SHIIIIIIIIIITT
SHEEERRRLOOOOOOCK
The funny thing about this "AI" griftosphere is how grifters will make some outlandish claim and then different grifters will "disprove" it. Plenty of grant/VC money for everybody.
-
What's hilarious/sad is the response to this article over on reddit's "singularity" sub, in which all the top comments are people who've obviously never got all the way through a research paper in their lives all trashing Apple and claiming their researchers don't understand AI or "reasoning". It's a weird cult.
