Cheating just gets more sophisticated
The hardest worker in academia today is ChatGPT. Over the last year, high school and college students turned in over 22 million papers that used some form of generative AI, according to data from Turnitin, which detects plagiarism and AI in schoolwork.
It’s been about a year and a half since generative AI tools like ChatGPT flooded the internet, and English departments were forced to add “bots” next to “SparkNotes” and “69 jokes” on the long list of things to look out for in personal essays. So, in response, Turnitin and other education companies created AI detection tools for teachers.
- Turnitin claims its detector has reviewed more than 200 million papers.
- Of those, 11% of papers may be 20% written by AI, while 3% of papers contain at least 80% AI-generated writing.
Educators are still hesitant to embrace AI detection
Experts say AI detection software isn’t as reliable as services that call out plagiarism—and false positives can harm a student’s academic career. The software itself may even contain bias about what it thinks is AI-written: A study last year found that seven AI-detecting tools had a 61% false positive rate when reviewing exams that test nonnative English speakers applying to colleges.
This study didn’t include Turnitin, which claims its false positives rate is around 1%. The company also stresses that its AI tool is not definitive.
Purdue University told instructors to be careful about accusing students of using AI with only Turnitin as proof, while other schools like Vanderbilt, Northwestern, and Montclair have told educators to stop using the AI detector entirely. The software itself has been described by administrators as a “black box” with no clear methodology for how it detects bot-written content.
Bottom line: AI regulation in the education system (like everywhere else) is a hodgepodge of vague rules that can’t quite keep up with how fast the technology is moving.—MM
