Artificial intelligence successfully passes middle school test in Germany this time

ChatGPT, which has been in our lives since 2022, is getting better with every new version. Even if students do not admit it, they continue to use this artificial intelligence for both their homework and their projects. According to studies, ChatGPT writes better essays than middle school students.

IT CAME FIRST BY A LARGE MARGIN

Scientists from the University of Passau tested new versions of ChatGPT in a test called the Bavarian Abitur. (The Bavarian Abitur is an exam that students take for middle school graduation, which resembles the high school entrance exams we have.) More than 100 educators who participated in the study graded both the students' papers and ChatGPT's paper. As a result, the chatbot came in first by a very large margin in terms of average score. 

In the study conducted on an English course, the 3.5 and 4 versions of the GPT language model that powers ChatGPT managed to achieve better results than students in all categories used by researchers for evaluation in grading, such as vocabulary variety, sentence complexity, nominalization, use of moods, and epistemic and discourse markers. While the students' average success grade was 3.9, ChatGPT 4 reached 5.25 points. 

PROBABILITY OF MAKING HUMAN ERROR IS VERY LOW

According to the teachers, the students' grades are not low, but ChatGPT is overly successful. Therefore, students are falling behind this generative language model. According to the researchers, ChatGPT's ability to learn is quite high, and the probability of it making human errors is also very low. 

Professor Steffen Herbold, who works at the University of Passau and led the research, stated in his explanation that he was very surprised by the results and emphasized that they did not expect such a clear difference. The researchers also announced that the necessary sensitivity towards artificial intelligence tools must now be shown in education. 

News Source: 12punto