AI catchers are easily bypassed and create a fake sense of security. It’s an arms race where the detectors get better at detecting, then the LLMs get better at not being detected, ad infinitum. Obsessing over them just distracts teachers from the real solution: redesigning assignments to test authentic human judgment and the tangible process of creation.
“We should be less preoccupied with catching cheaters and more concerned with whether our assessments are valid measures of learning… simply labelling a take-home task ‘No AI’ without redesigning its conditions, evidence, and criteria is largely unenforceable and risks devolving into performance theatre” (Rudolph et al., 2025).
References
Rudolph, J., Tang, F. X., Aspland, T., & Stafford, V. (2025). What Does `Good Teaching’ Mean in the AI Age? Journal of Applied Learning & Teaching, 8(2). https://doi.org/10.37074/jalt.2025.8.2.1