But they just don't happen to be doing the same thing. People claiming otherwise have to first prove that we are comparing the same thing.
This whole strand of “inteligence is just a compression” may be possible but it's just as likely (if not a massively more likely) that compression is just a small piece or even not at all how biological inteligence works.
In your analogy it's more like comparing modern calculator to a book. They might have same answers but calculator gets to them through completely different process. The process is the key part. I think more people would be excited by a calculator that only counts till 99 than a super massive book that has all the math results ever produced by the human kind.
They are doing "the same thing" only from the point of view of function, which only makes sense from the point of view of the thing utilizing this function (e.g. a clerical worker that needs to add numbers quickly).
Otherwise, if "the parts are all different, and the construction isn't even remotely similar", how can the thing they're doing be "the same"? More importantly, how is it possible to make useful inferences about one based on the other if that's the case?
The more you try to look into the LLM internals, the more similarities you find. Humanlike concepts, language-invariant circuits, abstract thinking, world models.
Mechanistic interpretability is struggling, of course. But what it found in the last 5 years is still enough to dispel a lot of the "LLMs are merely X" and "LLMs can't Y" myths - if you are up to date on the relevant research.
It's not just the outputs. The process is somewhat similar too. LLMs and humans both implement abstract thinking of some kind - much like calculators and arithmometers both implement addition.
Without a direct comparison to human internals (grounded in neurobiology, rather than intuition), it's hard to say how similar these similarities are, and if they're not simply a result of the transparency illusion (as Sydney Lamb defines it).
However, if you can point us to some specific reading on mechanistic interpretability that you think is relevant here, I would definitely appreciate it.
That's what I'm saying: there is no "direct comparison grounded in neurobiology" for most things, and for many things, there simply can't be one. For the same reason you can't compare gears and springs to silicon circuits 1:1. The low level components diverge too much.
Despite all that, the calculator and the arithmometer do the same things. If you can't go up an abstraction level and look past low level implementation details, then you'll remain blind to that fact forever.
What papers depends on what you're interested in. There's a lot of research - ranging from weird LLM capabilities and to exact operation of reverse engineered circuits.
There is no level of abstraction to go up sans context. Again, let me repeat myself as well: the calculator and the arithmometer do the same things -- from the point of view of the cleric that needs to add and subtract quickly. Otherwise they are simply two completely different objects. And we will have a hard time making correct inferences about how one works based only on how we know the other works, or, e.g. how calculating machines work.
What I'm interested in is evidence that supports that "The more you try to look into the LLM internals, the more similarities you find". Some pointers to specific books and papers will be very helpful.
> Otherwise they are simply two completely different objects.
That's where you're wrong. Both objects reflect the same mathematical operations in their structure.
Even if those were inscrutable alien artifacts to you, even if you knew nothing about who constructed them, how or why? If you studied them, you would be able to see the similarities laid bare.
Their inputs align, their outputs align. And if you dug deep enough? You would find that there are components in them that correspond to the same mathematical operations - even if the two are nothing alike in how exactly they implement them.
LLMs and human brains are "inscrutable alien artifacts" to us. Both are created by inhuman optimization pressures. Both you need to study to find out how they function. It's obvious, though, that their inputs align, and their outputs align. And the more you dig into internals?
I recommend taking a look at Anthropic's papers on SAE - sparse autoencoders. Which is a method that essentially takes the population coding hypothesis and runs with it. It attempts to crack the neural coding used by the LLM internally to pry interpretable features out of it. There are no "grandmother neurons" there - so you need elaborate methods to examine what kind of representations an LLM can learn to recognize and use in its functioning.
Anthropic's work is notable because they have not only managed to extract features that map to some amazingly high level concepts, but also prove causality - interfering with the neuron populations mapped out by SAE changes LLM's behaviors in predictable ways.
> That's where you're wrong. Both objects reflect the same mathematical operations in their structure.
This is missing the point by a country mile, I think.
All navel-gazing aside, understanding every bit of how an arithmometer works - hell, even being able to build one yourself - tells you absolutely nothing about how the Z80 chip in a TI-83 calculator actually works. Even if you take it down to individual components, there is zero real similarity between how a Leibniz wheel works and how a (full) adder circuit works. They are in fact fundamentally different machines that operate via fundamentally different principles.
The idea that similar functions must mean that they share significant similarities under the hood is senseless; you might as well argue that there are similarities to be found between a nuclear chain reaction and the flow of a river because they are both harnessed to spin turbines to generate electricity. It is a profoundly and quite frankly disturbingly incurious way for anyone who considers themself an "engineer" to approach the world.
In case you have missed it in the middle of the navel-gazing about abstraction, this all started with the comment "Please stop comparing these things to biological systems. They have very little in common."[0]
If you insist on continuing to miss the point even when told explicitly that the comment is referring to what's inside the box, not its interface, then be my guest. There isn't much of a sensible discussion about engineering to be had with someone who thinks that e.g. the sentence "Please stop comparing [nuclear reactors] to [coal power plants]. They have very little in common" can be countered with "but abstraction! they both produce electricity!".
For the record, I am not the one you have been replying to.
Sure, the parts are all different, and the construction isn't even remotely similar. They just happen to be doing the same thing.