Apple did the research; LLMs cannot do formal reasoning. Results change by as much as 10% if something as basic as the names change.

https://garymarcus.substack.com/p/llms-dont-do-formal-reasoning-and

LLMs don’t do formal reasoning - and that is a HUGE problem

Important new study from Apple

Marcus on AI
@ShadowJonathan not to sound antiintellectual, but isn't it kinda obvious that a *text* generator, no matter how complex, can't do abstract reasoning?
@halva @ShadowJonathan yeah, I appreciate the demonstrations, but this feels a little like, "New study confirms bicycles cannot fly."

@graue @halva @ShadowJonathan

Companies like OpenAI and their defenders claim generative AI can reason, learn, etc. We know it’s nonsense, but it’s still extremely important it gets called out.

@rubenerd @graue @halva @ShadowJonathan This is a huge problem. They are the experts. Hinton and Ilya claiming a function can have understanding. Why are they lying? Seems counter productive to scare the hell out of people. Well I know why Ilya is lying he just got $1bn...

@nf3xn @rubenerd @graue @halva @ShadowJonathan I doubt Hinton is lying although he’s probably wrong. There’s a problem in philosophy: is the mind separate from the body? If it’s not, then it should be possible to model the brain well enough to simulate thought processes (at least in principle.)

Computational physics tells us that there is a function that could perform the simulation and Hinton’s career is looking for it.

@MartyFouts @nf3xn @rubenerd @graue @halva @ShadowJonathan there’s a huge gap between “possible in principle” and “this does it now”. A kite can fly, but being able to build a kite doesn’t mean you can build an airplane