News for hackers. Stuff that matters.
Re:Does current AI represent a dead end? by legacynl on Friday December 27, 2024 @01:12PM (#42524509)
A little nitpick; a biological neuron is much more complex than it's ml-model equivalent. a simple weighted function cannot fully replicate a neuron.

That's why it's almost certain that a biological brain with a billion synapses outperforms a model with a billion parameters.

Re:Does current AI represent a dead end? by mort96 on Friday December 27, 2024 @01:15PM (#42524533)
Isn't that what they meant by this?

> the comparison is not apples-to-apples, because each synapse is much more complex than a single parameter in a weight matrix.

Re:Does current AI represent a dead end? by daveguy on Friday December 27, 2024 @01:50PM (#42524853)
It isn't just "not apples to apples". It's apples to supercomputers.
Re:Does current AI represent a dead end? by legacynl on Friday December 27, 2024 @02:17PM (#42525147)
well yeah but it's un-obviously a very big difference that basically invalidates any conclusion that you can make with this comparison.
Re:Does current AI represent a dead end? by mort96 on Friday December 27, 2024 @02:51PM (#42525447)
I don't think so: it seems reasonable to assume that biological neurons are strictly more powerful than "neural network" weights, so the fact that a human brain has 3 orders of magnitude more biological neurons than language models have weights tells that we should expect, as an extreme lower bound, 3 orders of magnitude difference.
Re:Does current AI represent a dead end? by joe_the_user on Friday December 27, 2024 @01:18PM (#42524557)
It's not a "nitpick", it's a complete refutation. LLM don't have a strong relationship to brains, they're just math/computer constructs.
Re:Does current AI represent a dead end? by Purplehermann on Sunday December 29, 2024 @01:58AM (#42538204)
The brain is insanely energy efficient, this is not the same as intelligence efficient