Plagiarism detection company Turnitin has released data claiming that at least 22 million papers submitted by students last year may have been written using generative AI, Wired reported.
Following the release of ChatGPT in late 2022 by OpenAI, millions of students have used similar chatbots to brainstorm ideas, inspire essays, and in some case write entire school papers. Schools have struggled to keep students in check and implemented the use of plagiarism-detection software to combat the rampant use.
Turnitin's technology has scanned more than 200 million papers written by mostly high school and college students. Of the documents reviewed, the software discovered 11% might contain artificial intelligence-written content in 20% of its content while 3% of the papers were found to have 80% or higher of AI produced content.
Turnitin states its writing detection technology is "is highly reliable and proficient in distinguishing between AI- and human-written text and is specialized for student writing."
Annie Chechitelli, chief product officer of Turnitin, told University Business in February that it will ultimately be up to student transparency and educator efficiency to mitigate the effects of AI in the classroom. She noted that teachers will likely be pushed to assign more challenging assignments.
"The interesting thing we'll soon be looking at is the unknown," she said. "Is AI going to start plagiarizing [itself]? At what point does it start repeating? When there is less human writing on the internet, what will that mean? I think this is still the beginning of understanding how AI generative writing becomes part of what people do and how we do it."
Chechitelli's comments appear prescient as AI might be soon used to evaluate its own work. The Turnitin report follows Tuesday's news that Texas plans to use AI chatbots to grade written exams by students with the implementation of their "automatic scoring engine." The state claims it will save $15 million to $20 million per year in human labor costs by using the technology.
While many educators are seeking a solution to the ubiquity of AI in the classroom, others are concerned the use of plagiarism detectors like Turnitin will lead to false accusations. Turnitin states its detector has a false positive rate of less than 1% when analyzing full documents.