Abstract: This talk discusses how students cheated in exams and other assessments in the run-up to Covid-19, then during Covid-19 in online exams, and how this landscape has changed with the advent of generative AI such as ChatGPT. We discuss how to prevent, detect and document various forms of cheating. These include contract cheating using external providers marketing themselves as "tutoring companies", impersonation techniques in unsupervised and electronically supervised exams, misrepresentation of personal circumstances to gain access to unsupervised assessments, communication in and out of supposedly secure exam rooms, and use of ChatGPT and other AI tools - all the way to students outsourcing their entire degree. Dealing with such problems calls for a concerted approach. This approach has a technological side: Academic networks and online platforms need to be secured, appropriate logging and real-time detection technology needs to be in place, back doors that allow students to obscure their helpers need to be shut, and technology to document and support the prosecution of cheats is a must-have. The talk introduces a number of initiatives we have taken with technology to make cheating harder and riskier for students. But there is also an organisational side: First of all, a need to upskill staff whose concept of academic misconduct is still too often that of a lone scholar who borrows half a sentence in Chapter 9 of their thesis without acknowledging the quote. Awareness of cheating techniques is key, as are efficient processes to deal with them. Finally, we look at the challenges that AI has brought, how to detect the use of ChatGPT & Co. and how this presents the academic sector with a huge question: How can we justify dedicating a large amount of human resources to the task of reading, grading and providing feedback on student submissions that are increasingly being generated by a machine, especially if the students in question have neither read nor understood the submission, and will never read the feedback?

​

​