My ex-wife doesn't have internal dialogue. Instead, when she is thinking through something she has images in her head of shapes morphing. She told me that she felt having an internal monologue would be horribly annoying and horribly slow.
LLMs might be a step towards AGI (assuming that's possible) but the prevailing belief that LLMs and human brains (the latter being the most complex structures in the universe that we know of) are already functionally equivalent in every way that matters seems more like an article of faith to me than fact.
this must be symbolic of something, nebulous, like how statistics can mean anything that will get more grant funding, but nothing actualy specific or informative.
we can see the lesser version of this in click title writing everywhere, so the whole thing is a war for attention, but of course people seem to form some kind of resistance or more often get sucked further and further in, but then become essentialy zero sum partisipants.
thats why the quest is to get them young and lock them into lifetime spending routine
Nice explainer. One nuance I’d be careful with: “the brain doesn’t think in words” is broader than what the study seems to establish. The stronger result is that the natural-language network is not necessary for the kinds of inductive and deductive reasoning tested here.
The AI implication is fascinating though. We tend to build systems where the language model is simultaneously the language interface, memory, planner, reasoner and controller. Human cognition may be pointing in the opposite architectural direction: separate persistent state, reasoning/control mechanisms and language, then let language be one tool among several.
That distinction seems more interesting to me than simply asking whether LLMs “reason in language.”
The study tested rule-induction puzzles (number sequences with hidden transformation rules) on two patients whose language networks were destroyed by stroke, plus fMRI on healthy volunteers solving the same puzzles silently. Both lines of evidence point the same way: the language network sits out during reasoning.
My ex-wife doesn't have internal dialogue. Instead, when she is thinking through something she has images in her head of shapes morphing. She told me that she felt having an internal monologue would be horribly annoying and horribly slow.
Did her brain get bricked every time she saw something of high enough quantity?
Neither do LLMs https://arxiv.org/pdf/2607.03502
Am I off base to view this as a strong hint AGI won't arrive through mere LLMs?
There have been strong hints since the 60s.
Probably not.
LLMs might be a step towards AGI (assuming that's possible) but the prevailing belief that LLMs and human brains (the latter being the most complex structures in the universe that we know of) are already functionally equivalent in every way that matters seems more like an article of faith to me than fact.
I made a 2-minute animated explainer of this study for a small channel I run: https://youtu.be/pzMBygLewZI?si=YBkRi8nP47RfTVhR — feedback on accuracy welcome.
this must be symbolic of something, nebulous, like how statistics can mean anything that will get more grant funding, but nothing actualy specific or informative. we can see the lesser version of this in click title writing everywhere, so the whole thing is a war for attention, but of course people seem to form some kind of resistance or more often get sucked further and further in, but then become essentialy zero sum partisipants. thats why the quest is to get them young and lock them into lifetime spending routine
Nice explainer. One nuance I’d be careful with: “the brain doesn’t think in words” is broader than what the study seems to establish. The stronger result is that the natural-language network is not necessary for the kinds of inductive and deductive reasoning tested here.
The AI implication is fascinating though. We tend to build systems where the language model is simultaneously the language interface, memory, planner, reasoner and controller. Human cognition may be pointing in the opposite architectural direction: separate persistent state, reasoning/control mechanisms and language, then let language be one tool among several.
That distinction seems more interesting to me than simply asking whether LLMs “reason in language.”
The study tested rule-induction puzzles (number sequences with hidden transformation rules) on two patients whose language networks were destroyed by stroke, plus fMRI on healthy volunteers solving the same puzzles silently. Both lines of evidence point the same way: the language network sits out during reasoning.