ChatGPT Surpasses University Students in Writing Assignments, Study Reveals

A recent study published in Scientific Reports has found that ChatGPT, an AI language model developed by OpenAI, is capable of achieving, and in some cases surpassing, the average grade of university students when answering assessment questions across various subjects. The research also discovered that a significant number of students would be willing to utilize ChatGPT for academic assistance, despite concerns from educators about plagiarism.

Researchers Talal Rahwan and Yasir Zaki conducted the study by inviting faculty members from New York University Abu Dhabi to provide three student submissions each for ten different assessment questions. ChatGPT was then tasked with generating three sets of answers for these questions, and the answers were assessed by three graders who were unaware of their source. The results showed that ChatGPT achieved an average grade that was equal to or higher than students in nine out of the 32 courses. Only mathematics and economics courses consistently outperformed ChatGPT.

The most noteworthy difference was observed in the “Introduction to Public Policy” course, where ChatGPT received an average grade of 9.56, compared to the average grade of 4.39 for students. These findings suggest that ChatGPT has the potential to offer valuable support in academic settings.

In addition to evaluating ChatGPT’s performance, the researchers also conducted a survey among 1,601 individuals from different countries, including students and educators. The survey revealed that 74% of students would consider using ChatGPT to assist with their assignments. However, the perception among educators was quite different. Across all countries, educators underestimated the number of students who planned to use ChatGPT and 70% of educators considered its use to be plagiarism.

Interestingly, the study also explored the ability of two AI text identification tools, GPTZero and AI text classifier, to detect AI-generated answers. It was found that both tools misclassified the ChatGPT-generated answers as human-written a significant percentage of the time.

These findings have important implications for the use of AI tools in educational settings. They highlight ChatGPT’s potential as a valuable resource for students while also underscoring the need for clear guidelines and policies regarding its use to address concerns of plagiarism.

Additional Information:
– The study was published in Scientific Reports and can be accessed [here](
– The research was conducted by Talal Rahwan and Yasir Zaki.
– The study was supported by Nature Publishing Group.

Source: – [link](