If the answer to “What is a mind?” is best approached by a quip like “Mind is more of a minding”, such that it is an activity rather than a thing, then I think the conditions of intersubstitutability are a “key” of sorts to understanding what that might mean.
Here I’m thinking that if a bundle of nerves and a bundle of circuits “amount to the same thing” then we must understand what the different bundles are doing in order to “amount to the same thing”: i.e. there are conditions wherein we accept that a bundle of circuits “does the same thing” as a bundle of nerves when talking about logical gates, for instance.
But those conditions are kind of begging the question on what a mind is – this isn’t bad, of course. Science is often made by begging the question and then testing to see what happens or comes from that assumption: guess and check, more or less, but more loose.
The is-ness I’d propose here are the conditions of intersubstitutability.
I don’t think we understand the neurons well enough to define such conditions at the moment. I suspect the present LLM’s are deficient. But, also, if we don’t have those conditions then we can’t claim the LLM’s are thinking (or, to be fair, not-thinking, at least on this basis – tho I think their outputs are fair examples of not-thinking )
This is just something that’s been bugging me as the world continues to adopt AI. I’ve been doing the round again recently of thinking about computers, animals, writing, humans, meaning and all that rot and this is what came of it to share amongst friends.
The substrate upon which thinking happens to occur may be irrelevant as long as the resulting processes I.e minding/thinking/consciousness are identical.
A = B if, for every sentence containing A, when we substitute B for A, the truth of that sentence does not change.
The parallel would be, perhaps, consciousness A is the same as consciousness B if, for every… what? Statement about A? Situation? That seems to need filling out.
If for every situation on which we say A is conscious, we find that when we substitute B, we say that B is conscious, then A = B?
My thought is that the conditions of intersubstitutability are such that if I arrange the wires and transistors in such-and-such a way then I can interpret them with respect to Truth Tables because they’ll operate the exact same way (and much faster than I formulate such conjectures).
So if a bundle of circuits does this or that with respect to logic its operation is intersubstitutable for if I had done the same brute logical force operation (due to the way we set up the circuits to operate like truth tables).
@Banno Excellent question. Though it seems a false dichotomy on reflection. I interpret and conclude. I could interpret the circuits as “only” doing the interpretation in order that I may conclude, or vice-versa.
Which I suppose still emphasizes the possibility of interpretation, though when I thought about it I could see the various “logical gates” as interpretation and the conclusion as “what’s relevant”, when I mean “interpretation” in a different way – the annoying frame from “the other side”, but hopefully not bound to consciousness :D.
Cognitive science often compares the mind to software and the brain to hardware. In this analogy, the brain is the physical computer, and the mind is the information-processing system running on it.
But beyond that, I get the feeling that you need much more than what you’ve been given, and you haven’t found it yet. And honestly, that’s understandable. Science still has enormous gaps in our understanding of the brain, and we’re nowhere near fully comprehending it. Any answer you get from an LLM is ultimately limited by the same incomplete scientific understanding we have as humans. It can explain what we currently know, but it cannot magically provide answers to questions we haven’t figured out ourselves.
This is me posting my most recent thoughts on this problem that I’ve posted about so many times before, with new words.
My random thoughts are meant to point out that these are not the same. Our incomplete and stupid understand we have as humans isn’t even in the same category as LLM’s. They are not, from what I can tell, intersubstitutable.
AI won’t replace human understanding. If we still don’t fully understand the mind, an LLM can’t simply give us definitive scientific answers about it. It can generate hypotheses and draw inferences from existing knowledge, but those outputs aren’t a substitute for genuine scientific discoveries. We still need to advance our scientific understanding as humans, or the information AI provides will remain limited by the gaps in our knowledge.
Mind as a physical process? I can’t help separating mind from matter, although without implicating substances. It doesn’t bother me to hear talk of “thinking machines” as though thinking is a physical process, but mind has more of a phenomenological angle to it. Physical and phenomenal. Thinking is something the physical brain does, and the phenomenolgy is what it “feels like” to be a brain. So the question for me is not whether LLMs think (they do), but whether they feel (I don’t know).
Mind is what consciousness uses to influence the world.
And influence is energy. So, it’s a sea of energy.
Energy with a hidden property that makes all of it part of the same system, makes it a part of the same stream of consciousness.
Born from the same source, mind stays bound to it. All parts of it.
And AI, it was designed to imitate what the mind does in the brain. When it goes there and uses strange methods to perform calculations to find whatever answer consciousness wanted to know.
)