June 27th 2024.
A recent study conducted in the UK has found that university exams written by artificial intelligence are nearly impossible to detect and can even outperform exams written by human students. This discovery has raised concerns about academic integrity in the education sector, as AI technology continues to advance and mimic human writing.
The study, which was published in PLOS ONE, involved 33 fake students from the University of Reading who submitted exam responses written entirely by an AI bot called ChatGPT-4. These responses were submitted for five different subjects in an undergraduate psychology degree. Surprisingly, the coordinators and markers of the exams were completely unaware of the experiment.
Out of all the exams submitted, only two were identified as forgeries by the markers. This means that the vast majority of AI-written essays went undetected. Even more impressive, the exam papers written by the chatbot received higher marks on average than those of real students. In fact, 83.4% of the AI-generated answers received higher grades than a randomly selected group of submissions from human students.
These results are a clear indication of the rapid advancements in AI technology, particularly in its ability to mimic human language and outsmart detection devices. The fact that the AI-written exams were virtually undetectable is a major concern for the academic community, as it raises questions about the validity and authenticity of student submissions.
The authors of the study highlight the fact that the use of AI in exams will only become more prevalent as the technology continues to improve. As its responses become more complex and abstract, it will become increasingly difficult for markers to distinguish between AI and human-written work. This is a worrying prospect for maintaining academic integrity.
The study also raises concerns about the widespread use of AI in exams, as it is impossible to estimate how many students may have used AI to complete their submissions. With the increasing media coverage of AI technology, it is likely that its use is already widespread among students.
In conclusion, the study's findings highlight the need for further discussion and measures to ensure academic integrity in the face of advancing AI technology. As the study authors note, the use of AI in exams is a complex issue that requires careful consideration and action to preserve the authenticity and credibility of academic work.
[This article has been trending online recently and has been generated with AI. Your feed is customized.]
[Generative AI is experimental.]
