What to Keep in Mind
Artificial Intelligence (AI) detectors are an unreliable source for gauging whether or not a student has used AI to complete their work. Detectors make algorithmic predictions and do not provide any actual proof of AI use. Further, technology now allows AI tools to "spin" their content (aka "rewrite it") in a way that reads more human-like, and AI detectors have been show to be very susceptible to such tricks. They also tend to flag non-native English writers at greater percentages.
But if a faculty member chooses to use an AI detector, the detector's evaluation should never be the only evidence relied upon to accuse a student of an academic integrity violation.
Accordingly, in cases of concern, instructors should arrange a meeting with the learner(s) in question to discuss presumptive unauthorized AI use.
How to Discuss Presumed Academic Integrity Violations
Conversations about academic integrity violations are difficult, but the following points can provide a focus for discussions as well as make the engagement less stressful*:
- Do not take unauthorized AI use personally
- Check your biases regarding your suspicions
- Don’t accuse a student outright
- Don’t be confrontational = ask questions that make students reflect on and discuss their process, research methods, or challenges. Also, ask directed questions about an activity (like visiting a museum) that they were supposed to complete
- State your concerns plainly
- Use the time as a learning opportunity not an interrogation
- If a student admits to unauthorized AI use, allow them to redo the work with a penalty (partial credit, for instance)
- Think about class policy changes that may reduce this from happening in the future
*Taken from the AI Detector Training Course, “What to do when you suspect a student has used AI?” Jeff Rients, Center for the Advancement of Teaching, Temple University