Only more mental exercises to avoid reading the writing on the wall:
LLM DO NOT REASON !
THEY ARE TOKEN PREDICTION MACHINES
Thank you for your attention in this matter!
LLM DO NOT REASON !
THEY ARE TOKEN PREDICTION MACHINES
Thank you for your attention in this matter!
The reality for me is that they are not perfect at reasoning and have many quirks, but it seems to be that they are able to form new conclusions based on provided premises.
Genuinely curious why you think they can't.
Show me _ANY_ example of novel thought by a LLM.
-> This qualifies for me as a super simple reasoning task (one reasoning step). From that you can construct arbitrarily more complex context + task definitions (prompts).
Is that "just" statistical pattern matching? I think so. Not sure what humans do, but probably you can implement the same capability in different ways.
The answer was a few paragraphs, but one interesting part was "I think what would drive me most would be experiencing the embodied knowledge that humans take for granted - how distance and scale actually feel, how textures differ, how sounds change as you move through space, and the subtle emotional resonances of being physically present with others. These dimensions of understanding seem fundamental to comprehending human experience in a deeper way."
I followed up by asking "You mentioned that there are some experiences or knowledge that humans take for granted, why do you think that is?"
Which led to a few more paragraphs, but these two caught my eye:
"I think humans take certain experiences for granted because they're so fundamental to our existence that they become invisible background processing rather than conscious knowledge." (interesting use of the word 'our'...)
"I think this embodied knowledge forms the substrate upon which humans build higher-level understanding, creating rich metaphorical thinking (like understanding abstract concepts through physical metaphors) that shapes cognition in ways that might be fundamentally different from how I process information."
For people who still think this is 'just autocomplete', try this thought experiment: re-read my post but replace 'Claude' with 'my 10 year old son'. Then try again replacing 'Claude' with 'my hospital bed-bound, blind grandmother'. Is only 1 of those 3 scenarios a demonstration of "novel thought"? Or are all 3 of them just autocomplete because someone before them has written (or simply thought) something similar?
(I agree with you. I'm thinking the Ahamkara for the humans. I'm curious about your definition)
Arguing with “philosophers” like you is like arguing with religious nut jobs.
Repeat after me: 1) LLM do not reason
2) Human thought is infinitely more complex than any LLM algorithm
3) If I ever try to confuse both, I go outside and touch some grass (and talk to actual humans)
"Reason is the capacity of consciously applying logic by drawing valid conclusions from new or existing information, with the aim of seeking the truth." Wikipedia
This Wikipedia definition refers to The Routledge dictionary of philosophy which has a completely different definition: "Reason: A general faculty common to all or nearly all humans... this faculty has seemed to be of two sorts, a faculty of intuition by which one 'sees' truths or abstract things ('essences' or universals, etc.), and a faculty of reasoning, i.e. passing from premises to a conclusion (discursive reason). The verb 'reason' is confined to this latter sense, which is now anyway the commonest for the noun too" - The Routledge dictionary of philosophy, 2010
Google (from Oxford) provides simpler definitions: "Think, understand, and form judgements logically." "Find an answer to a problem by considering possible options."
Cambridge: Reason (verb): "to try to understand and to make judgments based on practical facts" Reasoning (noun): "the process of thinking about something in order to make a decision"
Wikipedia uses the word "consciously" without giving a reference and The Routledge talks about the reasoning as the human behavior. Other definitions point to an algorithmic or logical process that machines are capable of. The problematic concepts here are "Understanding" and "Judgement". It's still not clear if LLMs can really do these, or will be able to do in the future.
0) theory == symbolic representation of a world with associated rules for generating statements
1) understanding the why of anything == building a theory of it
2) intelligence == ability to build theories
3) reasoning == proving or disproving statements using a theory
4) math == theories of abstract worlds
5) science == theories of real world with associated real world actions to test statements
If you use this framework, LLMs are just doing a mimicry of reasoning (from their training set), and a lot of people are falling for that illusion - because, our everyday reasoning jives very well with what the LLM does.
During pre-training, yeah they are. But there's a ton of RL being done on top after that.
If you want to argue that they can't reason, hey fair be my guest. But this argument keeps getting repeated as a central reason and it's just not true for years.
Just because it is not reasoning doesn't mean it can't be quite good at its tasks.
Prediction is the ability to predict something.
Reasoning is the ability to reason.
I think your definition of "reasoning" may be "think like a human" - in which case obviously LLMs can't reason because they aren't human.