The performance of large-scale language model (LLM)-based artificial intelligence has improved day by day, and the most recent version of ChatGPT, o1-Preview, has risen to the level of getting all but one question right in the Korean language section of the CSAT this year.
The previous ChatGPT models, o1-Mini and GPT-4o, were significantly different from those in the 4th grade of 78 and 75 respectively.The question with
o1-preview wrong was a non-literary fingerprint No. 8, and the actual test-taker incorrect answer rate reached 81.5%.
Mark AI, a startup that compares scores in the Korean language area of the CSAT by artificial intelligence model, said o1-preview has been confirmed to have the language level and logic of the top 4%.
At the same time, he predicted that open-source AI will develop at a rapid pace and that the prelude to the end of the white collar will be opened soon.
※ 'Your report becomes news'
[Kakao Talk] YTN Search and Add Channel
[Phone] 02-398-8585
[Mail] social@ytn.co.kr
[Copyright holder (c) YTN Unauthorized reproduction, redistribution and use of AI data prohibited]