
Uncovering ChatGPT's Logical Abilities
4.0 Solutions Podcast
00:00
GPT 3.5 and GPT 4: A Comparison
GPT 3.5 got 47 of the 80 questions correct, which is score of 58.75. With GPT 4, it went to 70% what's very important because 70% is my historical number to be allowed to enter at the senior level. So we got 70% of section one right, 90% of section two right, only one question in section three right. Not even a fifth chance of a second.
Play episode from 17:19
Transcript


