[ACCount37]

First, fix your formatting. It's a fucking mess.

Second, what is the difference? Is it that one thing has an immortal soul, and thus Actual Intelligence and Actual Reasoning and Actual Learning, and the other has no soul, and a Pale Imitation of Intelligence, At Best?

Because I've seen versions of this "it's not actually thinking" for actual fucking years, and the difference between "actually thinking" and "not actually thinking" always seems to boil down to "I don't want LLMs to be actually thinking, so I will bend the definitions and twist the qualifiers and move the goalposts until they aren't". No one ever made an ActualThinkingBenchmark on which humans score 100% and LLMs score 0%.

Nothing but human insecurity, in my eyes. There was never a principled difference. Just a way to operationalize some "I'm unique and special and better than a matrix math machine" vibes.

[f_klem]

Agreed, formatting was kind of f, but there is no need to be rude.

I wasn't saying there was any difference. All I'm saying is that the claimings the AI research field does are based on false assumptions. And from false assumptions, you cannot reach a proper conclusion.

Whether an AI system can reason and think like if it where a human being, or not, I don't care. I'm fine with either: it is just technological advance. But making claims based on false assumptions, and then being fooled by how 'wonderful' or 'spectacular' the results are, is, at least, naive.

> Nothing but human insecurity, in my eyes. There was never a principled difference. Just a way to operationalize some "I'm unique and special and better than a matrix math machine" vibes.

This is just something I don't get. People ignorant of technique are insecure and afraid. People that know how technology works, and thus investigate and know how it works fundamentally*, were never afraid or insecure.

[ACCount37]

A lot of people who "know how technology works" just went looking for copium, and found some. Now, they "know" a comforting lie - something like "it's just next token prediction".

Very comforting, that, but actively harmful to understanding.

The understanding starts with: we don't actually know how LLMs do what they do. They're more grown than designed. And it only gets worse from there. Very little comfort to be found in modern AI.

[f_klem]

There are two things here: one is how an LLM is fundamentally structured and designed, the other is how an LLM distributes and 'lays out' the relationship between inputs and outputs through layers and weights.

We might not know how the actual distribution works, but we do know how it i s fundamentally structured and designed -- because we did it. We also know that there is something like a representation system inside them. And we also know that human beings do not hold 'internal representations' like any AI system needs to. So there isn't any 'intrinsically magical' in modern AI systems.

[ACCount37]

And knowing that structure is about as meaningful as knowing "a PC consists of a keyboard, on which you type, a screen, at which you look, and a processor, which does things with binary logic".

None of that helps you understand how exactly LLMs do what they do. Because it describes an interface, not a mechanism.

The inner mechanisms of an LLM are more learned than designed. We know what an LLM does on a low level, but going from that to understanding how they work is like trying to understand how a web browser works by looking at netlists of a CPU. Low level understanding does not grant you high level understanding for free.

But ignoring all of that lets you cling to a very comforting "we understand LLMs because we made them". Ha ha. As if.

> And we also know that human beings do not hold 'internal representations' like any AI system needs to.

Bold fucking claim. Got a source on that?

Because neurobiology has been trying to crack neural representations - the very internal representations brains use - for as long as it existed, and with some success. Both reading and injecting internal representations into the brain is possible now, in narrow cases. The specifics vary region to region, but sparse population coding is a true staple. Today's SOTA for wrangling this mess is ML decoders, and not by a coincidence.

[f_klem]

We know how LLMs learn at the fundamental level. What we do not know is the actual dynamic process of encoding embeddings and their distributions.

Your analogies about the PC and web browser are not correctly formulated, because in the case of the PC you talk about 'external components' (you should be talking about cpu arch, structure, digital components, interfaces, etc); in the case of the web browser, you should be talking about modules, code, etc.

We do know how LLMs are laid out: layers, att heads, etc. So what we need to look at are the fundamental possibilities of the structure of LLMs, not how the weights are distributed.

> > And we also know that human beings do not hold 'internal representations' like any AI system needs to.

> Bold fucking claim. Got a source on that?

Part of the sources are in the books I mentioned. Nonetheless, you can still fact-check and refute in an adult and serious manner, not in an disrespectful and arrogant way. If my claim sounded arrogant I apologize, but then as I already mentioned, my references back that claim.

Regarding internal representations in the brain: I guess you are referring to areas of the brain being activated when a subject receives a stimuli, and this is tested through MRI. I would be cautious to causally relate stimuli to neuron activations, since you first need to know if the exact configuration of cell involved and their connections allow for such representation (which I think it is still not known -- again, AFAIK, the contrary seems to be the case).

[ACCount37]

Your references that "back that claim", which are in "books you mentioned", which you "mentioned" who knows where.

Yeah, no. I'm not walking that chain. If you want to, do it, but for now, I'm filing it as "has no evidence and knows it".

By now, there's plenty of works, up to and including direct neural interfaces. Utah arrays, Michigan arrays. Stab the brain, dump the spike trains, decode. You crack the manifold open by correlating to known stimuli using ML, and generalize from there to unknown stimuli. There is no need to "know the exact configuration", and few bother - you put your hardware into the part of the brain you want (top level map is consistent enough brain to brain), gather a set of reference points, and use them to anchor the rest of the decoding process.

Why use ML? Because you need a very expressive correlator to bridge the gap between known inputs and the products of whatever transformations the brain subjects them to before they show up in spike trains.

> So what we need to look at are the fundamental possibilities of the structure of LLMs, not how the weights are distributed.

And the fundamental possibilities are... what exactly? We know the I/O planes, we know the possible flow of information, now, what does that give us?

We know enough to prove that a transformer LLM can implement a Turing machine, the same way a CPU can implement a Turing machine. So an LLM is capable of performing arbitrary computation within its capacity. That's it. That's the upper bound.

What follows is: if you can represent "thinking" as a computational process, you can implement it with a Turing machine, and thus, an LLM can be made to think. That proves LLMs can think. But not that the existing ones do or don't! Because that's the entire thing about upper bounds!

We've looked at LLM architecture, and learned basically nothing about whether LLMs think, other than "it's not impossible". That's the actual "fundamental possibilities" we derived from knowing the architecture. One step above worthless. Oh fun.

(If thinking requires hypercomputation, then, nope. LLMs are out. Good luck proving that it does though.)