ChatGPT Scores Highest in Japan's Top University Entrance Exams, Outperforming Human Top Scorers

Recently, ChatGPT achieved the highest scores in the entrance exams of Japan's University of Tokyo and Kyoto University, outperforming human top scorers. According to LifePrompt's test, the ChatGPT 5.2 model scored 50 points higher than the human top scorer in the natural science (medicine) exam of the University of Tokyo, with a full mark in math; it also exceeded the highest admission scores in Kyoto University's law and medicine exams.

Last year, the model failed to pass all entrance exams of the University of Tokyo, but this year it made a breakthrough. During the test, exam questions were converted into image data and input into the model, and essay questions were graded by teachers from Kawaijuku. However, ChatGPT's score rate in essay questions like world history was only 25%, while its English score rate reached 90%.

Experts point out that AI excels at data induction and should not compete with humans on the same track. The current college entrance exam model focusing on knowledge memory needs to be re-examined and reformed.