© 2025 WNIJ and WNIU
Northern Public Radio
801 N 1st St.
DeKalb, IL 60115
815-753-9000
Northern Public Radio
Play Live Radio
Next Up:
0:00
0:00
0:00 0:00
Available On Air Stations

Perspective: The myth of AI detection

AI prompt: A robot teacher grades student papers.
Pixlr AI
AI prompt: A robot teacher grades student papers.

If all you want for Christmas is an AI detector, I’m sorry to say that you are going to be sorely disappointed. Let me explain.

 

Since 30 November 2022—the day that OpenAI released ChatGPT on the world—educators and school administrators have been desperately wanting and waiting for an AI detector. If students were going to use these new large language models to do their homework, then the least we could do is develop an AI tool that could detect this AI generated content. This is a classic fighting fire with fire strategy.

 

But don’t get your hopes up. As of today—a good two years since the introduction of what some have called the most perfect plagiarism tool ever invented—there is no credible way to detect AI generated written material. Thus, is it virtually impossible for teachers to differentiate a machine generated term paper from that produced by an actual human being. There may be tells, and we could perhaps delve into them. And incidentally the use of the verb “delve” had been identified as one of these tell-tale signs.

 

But none of these distinguishing characteristics have amounted to much. To make matters worse, existing prototypes of these silver-bullet solutions have routinely identified material written by students for whom English is a second language as something that has been generated by AI, thus putting an already vulnerable population of young people under additional suspicion and scrutiny.

 

So far, AI detection is a pipe dream. It is, I am sorry to say, just as real as Santa Claus.

Northern Illinois University professor and author