345
When billion-dollar AIs break down over puzzles a child can do, it's time to rethink the hype
(www.theguardian.com)
this post was submitted on 12 Jun 2025
345 points (97.0% liked)
Fuck AI
3561 readers
849 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Cars are not marketed and expected to fly.
Oh, and Large Language Models say where exactly, that they are good at solving mathematical problems?
Here's Microsoft advertising Copilot "makes solving maths problems a breeze" with an entire how-to article.
https://www.microsoft.com/en-us/microsoft-copilot/for-individuals/do-more-with-ai/learning-and-education/how-to-use-ai-for-math-calculations
That’s not the LLM solving the problems. The LLM understands the user request and has the ability to use a math solver. Additionally: “You can also integrate Copilot with the Wolfram Alpha plug-in to gain access to vast computational, mathematical, and scientific knowledge.”
Cool, and anyone can just access Wolfram Alpha directly - as they've been able to for like 15 years - and make fewer mistakes than they would by using it via an LLM.
The LLM (Copilot) is being advertised as a maths tutor. It's not good at maths. It doesnt understand anything about maths. All it can do is send your query through to Wolfram Alpha and then spit it back to you.
Hype.
Some of the tested models were specifically "reasoning" LLM models. Please tell me that the "reasoning" model is not intended to be used for "reasoning". Please.