Key takeaways:
1. The unauthorised use of AI in assessment must be treated as fraud and reported to the EB.
2. The EB requires sufficient evidence to rule fraud.
3. Some BUT NOT ALL EBs advise teachers to conduct an oral inspection to collect further evidence.
4. Since identifying the unauthorised use of AI is becoming increasingly difficult, work on designing AI-resilient assessment.
5. For questions and support, contact: f.m.frittella@utwente.nl

Any unauthorised and/or undeclared use of AI tools must be treated as academic misconduct, more specifically, as fraud. As such, the case must be reported to the programme’s Examination Board. As a first step, please consult your EB to find out about their procedures you should follow.

Note that to act upon the reported case, the EB needs sufficient evidence that AI was used.

What constitutes evidence for AI use?

AI detectors report (if officially available at UT and GDPR compliant): At present, such tools cannot reliably detect AI use; hence, the reports cannot be regarded as undeniable proof. However, they may constitute part of the evidence presented by teachers for suspected AI use. Note that the UT may not have a contract with an AI detection tool, and teachers are advised against using additional detectors since these tools may not be GDPR compliant.
Signs in the assessment product: Teachers may use several aspects of the assessment product as grounds for their suspicion that AI was used. These signs include elements typical of AI tools, such as conversational elements in students’ text (“certainly!”), citation of non-existing sources, (in)consistency in style and lengthy but “contentless” elaborations typical of ChatGPT. Mastery of the content exceeding what was offered in the course may also be a sign of AI use.
Follow-up oral inspection (IN SOME EBs ONLY): Some but not all EBs advise their teachers to conduct an oral inspection with the student to collect further evidence before reporting the case to the EB. This is an intermediate step to collect further evidence for the EB before reporting the case; it complements and does not replace the EB’s own investigation. Since the EB needs sufficient evidence to rule on an (AI) fraud cause, conducting a follow-up oral inspection with students is recommended for teachers to collect evidence about the suspicion of AI use, particularly focusing on the assignment content, which the EB cannot verify otherwise. In the inspection, teachers can test whether students’ mastery of the content corresponds to that of the assessment product, whether students can explain how they came up with aspects of the product and inquire about specific signs of AI use in the product. Consult your EB to find out whether they advise this practice, in which case you may use the guidelines below. Note that if you wish to use this instrument, you must specify it upfront in your course AI policy as a possible consequence faced by students in case of suspicions of unauthorised AI use.

📄 Download our guidelines on how to conduct and report the outcomes of an oral inspection. Consult your EB before using these guidelines as they may use different procedures and guidelines.

Please, note that:

If the evidence is insufficient, the EB cannot rule fraud. However, if the assessment is poor, this may still impact the grade and lead to the student failing the assignment.

Since it is increasingly difficult to recognise AI-generated outputs, we advise teachers to examine their assessment methods critically. You may also consider various measures to ensure the reliability of your assessment in times of AI. If you need support with redesigning or tweaking your assessment, contact: f.m.frittella@utwente.nl