OpenAI, which has not gone off the agenda with its artificial intelligence tools, has recently launched its most advanced language model, GPT-4. The new model offered for ChatGPT Plus subscribers amazed everyone with what it could do.
Now Geek.com.tr has measured how GPT-4 will perform in a different field with a study it has done and once again showed us the development of artificial intelligence. The language model was used in the study, which is one of the exams taken by those preparing for university in our country. Basic Proficiency Test (TYT) inserted.
GPT-4 outperformed 99.50% of TYT participants

Geek.com has enabled the language model to solve the 2021 YKS TYT exam through the ChatGPT Plus plan. The artificial intelligence tool answered 40 Turkish, 20 social science, 20 natural science and 40 basic math questions in the exam. The model also explained how she answered the questions.
geek. com
As a result, in the Turkish part of the model there were 30 correct and 9 incorrect answers. 27.75 net18 true and 1 false in social science questions 18.75 net (1 question was considered correct because it contained images that could not be modified for the model).
geek. com
On the other hand, in Science section, there are 12 right and 4 wrong answers. 13.75 net (4 questions with pictures counted as 3 right and 1 wrong), and finally 18 correct and 14 wrong answers in math questions. 17.50 net (8 questions with pictures were counted as 4 correct and 4 incorrect).
geek. com
As a result of all these, GPT-4 achieved 419 points of the 2021 TYT exam. As a result of the examination of ÖSYM’s data, the model was determined by the number of people who took the exam in Turkey in that year with this score. Better than 99.50% and may be in the top 10,000 showed. Let’s add that about 2 million 416 thousand people took the exam.
If you want to take a closer look at the nerd’s research to the link here you can click.
Affordable exchange campaign for those who want to renew their Apple computer