Artificial intelligence and essay writing are in the news once again, but this time, it’s the detection software that’s getting all the attention. An AI detector recently concluded that the US Constitution was very likely written by artificial intelligence. I’m not sure whether the Founding Fathers would feel complimented or embarrassed. But the results raise an important question: when AI is used to detect cheating, what happens when the AI is wrong?
AI detectors were supposed to be the antidote to ChatGPT-written college essays, helping keep to integrity of these admissions criteria alive. Now, it appears the detectors are neither good nor accurate enough (yet, anyway) to serve that purpose. The question of what happens to students whose essays get incorrectly flagged has not been answered.