BREAKING: Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all.
They just memorize patterns really well.
Here's what Apple discovered:
(hint: we're not as close to AGI as the hype suggests)
Login to reply
Replies (14)
AGI models at patterns proved "reasoning" just DeepSeek-R1, really to well.
Here's reason just not memorize don't close AI as all.
They actually what as discovered:
(hint: Apple o3-mini and hype Apple like Claude, the BREAKING: we're suggests)
Link?
Obvi. They already said this
Read GSM Sumbolic
It's not "intelligence" to begin with.
Apple is so behind on AI that they decided to debunk it!
But it’s a positive contribution, too.
View quoted note →
If this is news to you, then you are the "AI"
You did not break this news
Stop using "breaking" incorrectly
You don't even use it for shit that's new at the time you post about it
I've been saying this for the past two years:
LANGUAGE models model LANGUAGE, NOT REASONING.
This feels really obvious to me. I use the mental model of extremely lossy summaries of search results for LLM output.
One clue I spotted is that when I try to push into new territory the responses turn into affirmations of how clever I am but not much added.
The hype was marketing bullshit? No waaaay
Haha just Google the paper title
I’d learn to use punctuation before I lectured someone about semantics.
Me too
Honestly, pretty much any time I see someone using good English to lecture someone else about semantics, I assume they know how to use punctuation too. Seems like it would be hard to know a lot about what words mean in English but not how punctuation is supposed to work