A university student has recently been caught using ChatGPT. 

The student, from the University of Bolton, referenced a ‘leadership book about witches and wizards in their essay’, as well as referencing a dissertation from 1952 and a journal article from 1957. Upon being caught by the University of Bolton’s standards and enhancement office, the student failed their essay and was forced to resit.  

This is the first known student that has been caught using ChatGPT to cheat in an assessment.

ChatGPT is an artificial intelligence (AI) chatbot developed by OpenAI and first accessible in November 2022. It is a ‘sibling model’ to InstructGPT, which is trained to follow a prompt instruction and provide a detailed response in return.

ChatGPT has limited knowledge after September 2021, but OpenAI’s more advanced GPT-4 has been available for students at £16 a month to subscribe to ChatGPT Plus; the more advanced Chatbot would have been able to cite more recent information, such as the student’s use of Simha’s ‘Leadership Insights for Wizards and Witches’.

The university said there were four reasons why they worked out that this student had used AI. They said ‘a number of statements in section links and in the practice section do not make sense.’ Additionally, the essay only referenced the two journals which were available to students on the university’s integrated system, Moodle, in which students can access all their course materials; the student used no sources from the module’s reading list. The university staff investigating the matter described the rest of the material in the essay as ‘obscure references’. 

Universities are increasingly questioning the reliability of Turnitin’s AI detection software. The digital plagiarism detection service is used by more than 10,000 higher education institutions around the world and almost every university in the UK. Turnitin provides a similarity report of the student’s work against their database and every previous essay submitted using the service. It shows universities how many phrases and sentences used by the student have appeared elsewhere. However, AI chatbots pose a problem in that they provide unique answers to the user each time. Recently, Turnitin addressed these concerns by launching a detection software that they claimed could show how much of an essay was produced by AI; the company boasted that they could ‘identify AI-generated text with 98 percent confidence’. 

A couple of months ago, The Tab reported that students and academics had visited the ChatGPT website more than one million times in just two months. There were 128,402 recorded visits from eight Russell Group universities in December 2022 and 982,809 in January. Warwick had the highest number of visits in this period with 850,141, falling around the university’s assessment period, and these visits were using the university’s Wifi.

Read More News Here: