Elektrine lite

← Feed

@lipsumdev@mathstodon.xyz

2026-09-11 22:21 UTC

@antoinechambertloir@mathstodon.xyz > I see no argument for which automatic generation of arguments, possibly looking in various directions at the same time, has anything to do with the stunning phenomenon called intelligence that we, humans, are used to experiment I will not pretend to understand this stunning phenomenon myself but, you might want to have a look at a recent computational neuroscience book like this one https://neuronaldynamics.epfl.ch/ Neuroscientists have been studying individual neurons for a long time now (and the book describes these models in length). Going from an understanding of the individual neuron to an understanding of billions of them connected is a giant and difficult leap. The book describes one approach that I find rather illuminating (https://neuronaldynamics.epfl.ch/online/Ch17.S2.html), which is due to Hopfield (he got the 2024 Physics Nobel prize for that). I think these (Hopfield networks) are a good introduction to energy-based models and (biological) neural networks. In some sense, I think it gives some intuition on how things can emerge at large scale from local interactions and perceptual senses. (Regarding the relation between logic and energy-based interactions, I also tried to talk about it on my personal blog here https://lipsum.dev/2025-01-1-sat-frustration/). (Now, that is not how LLM work, but that is another topic.)

Replies (1)

  • @lipsumdev@mathstodon.xyz 2026-09-12 05:52

    @antoinechambertloir@mathstodon.xyz I will just add a few words regarding that last point and LLM (or Artificial NN in general), because I think that's highly relevant: There is a principle in physics which is the principle of locality : information does travel from neighbour to neighbour. Current Artificial Neural Network technologies do *not* respect that locality principle (because of how backpropagation works). That is a major difference between the architecture of Artificial NN and Biological NN (of their conceptual models, I mean). That is something developed in https://www.igb.uci.edu/~pfbaldi/book-deep-learning-in-science/Ch7-LocalLearnihng.pdf. --- That being said, I wouldn't exclude that Artificial NN can perform as well as humans at just because their architecture is different from (the models of) our own. I have been quite impressed by the way Claude Code performs (it's impressive, reallly). As you mention, there is a similarity between mathematics and programming. This similarity presupposes that the relevant mathematical concepts have been formalized in some sense (I mean, that the semantics living in the brain of the mathematicians has been put on paper), and that's where I would have questions regarding their capabilities and their ultimate limitations; however, it's hard to exclude anything at this point (at least, for me, with the information I have and may process with my limited own capabilities :-) ). Anyway, there is something which I find annoying in that whole thing: how trust can be established ? I think that is mainly a human problem.

    Open ##5153713