Cheating Detector: Students Submitted 22 Million Papers Written by AI Last Year
Students submitted more than 22 million papers last year that were likely written by AI tools, according to new data published by Turnitin, a software service that checks papers for plagiarism.
Turnitin says its plagiarism detection tool found millions of papers that may include a significant amount of AI-generated content, according to a report by Wired.
Last year, the company launched an AI writing detection tool and trained it on a slew of papers written by high school and college students, as well as known AI-generated content. Turnitin says its detector tool now has a false positive rate of less than one percent when examining entire documents.
Turnitin noted that its detector tool has analyzed more than 200 million student-submitted papers and found that 11 percent of them appear to be at least 20 percent made up of AI-generated written language, with three percent of the papers being at least 80 percent written by AI.
These types of tools like OpenAI’s ChatGPT have become a major problem in the world of academia, as students are increasingly using the tool as their go-to source for cheating.
As Breitbart News reported, a study published last year found that 17 percent of students at Stanford University admitted to using ChatGPT on their final exams. Students at an elite academic program at a Florida high school were accused of cheating by using ChatGPT to write their essays.
While educators want to hold students accountable for cheating, accurately detecting the use of AI in writing assignments can be difficult.
Notably, AI detection tools can pose a risk of false positives against English language learners.
As Breitbart News reported, a Stanford study found last year that AI detection tools are biased against non-native English speakers and falsely accuse them of cheating.
Turnitin claims it has trained its plagiarism detector on writing styles of both English language learners and native English speakers, and the software company has been listed as one of the most accurate AI language detectors, according to an October study, Wired reported.
Annie Chechitelli, Turnitin’s chief product officer, suggested the software be used as a way for teachers to start a conversation with their students abut the nuance in using generative AI, rather than as a means for passing or failing a student.
“People don’t really know where that line should be,” she said.
-Breitbart
