ChatGPT could be a top performer in university assignments.

In a new study, ChatGPT has shown the potential to surpass university students in various academic fields. 

Researchers from the UAE conducted an eye-opening investigation, demonstrating ChatGPT's abilities across subjects like computer science, political studies, engineering, and psychology. 

The research, led by Talal Rahwan and Yasir Zaki at New York University Abu Dhabi (NYUAD), embarked on a mission to assess ChatGPT's academic mettle. 

They enlisted faculty members teaching 32 diverse courses, each providing three student submissions for ten assessment questions in their respective domains. 

ChatGPT was then put to the test, generating answers to the same questions. 

Independent graders, unaware of the source, meticulously evaluated the responses. 

The outcome was striking: ChatGPT achieved an equivalent or higher average grade than students in 9 out of 32 courses. 

Only in mathematics and economics did students maintain a slight edge.

The AI was strongest in an 'Introduction to Public Policy' course, which saw ChatGPT's average grade hit 9.56, while students lagged behind at 4.39.

Surprisingly, when 1,601 individuals from Brazil, India, Japan, the US, and the UK were surveyed about ChatGPT's role in academic assignments, 74 per cent of students expressed their willingness to enlist its help. 

However, educators, perhaps due to concerns of plagiarism, underestimated this figure, with 70 per cent suggesting they would consider its use as dishonest.

The research also spotted a significant challenge. 

Two AI text identification tools, GPTZero and AI text classifier, struggled to differentiate between ChatGPT's responses and those crafted by humans, misclassifying the AI-generated text as human 32 per cent and 49 per cent of the time, respectively.

The full study is accessible here.