"Researchers at the University of Pennsylvania found that Turkish high school students who had access to ChatGPT while doing practice math problems did worse on a math test compared with students who didn’t have access to ChatGPT." hechingerreport.org/kids-chatgpt...
Researchers compare math progress of almost 1,000 high school students
Good to see you around here!! I thought I wouldn’t read you anymore after twitter was blocked in Brazil.
“open book test” and the book is the necronomicon
So AI is not smart enough to know when it has the wrong answer to a HS math problem. Neither are the VC investors who believe that AI is going to make them mega wealthy by replacing every creative being on the planet. Do the math...oh wait... 😇
You really do just link the most fascinating things! This is a great paper and honestly, I’d never heard of this publication before. It seems great too. The authors of this paper are very humble and careful to make huge leaps on the implications of their paper but the findings are striking
It just needs more training, bro. Just a little more training on copyrighted materials and it will be good, bro. For real, bro.
...innnnteresting. I know it was toxic for English and History, but how's it make Math worse?
Yes, *language* models are bad at math.
I wonder what the results would have been if the students also had access to ChatGPT DURING the test.
LOL Who thought otherwise, Dr. Birhane?
"Those with ChatGPT solved 48 percent more of the practice problems correctly, but they ultimately scored 17 percent worse on a test of the topic that the students were learning."