Plagiarism checker finds AI-generated content in 22M student papers

In the past year, students have handed in more than 22 million papers that might have used generative AI programs, per plagiarism detection company Turnitin.

Turnitin launched its own AI writing detection tool a year ago. Since then it has analyzed 200 million papers, primarily from high school and college students.

It found that approximately 11% of the papers showed 20% traces of AI-generated content and 3% consisted of 80% or more AI writing.

With such widespread use, educators face challenges in detecting AI use in student papers. Several have resorted to using AI tools in their grading processes.

AI detection tools aren’t reliable
There are concerns about biases in AI detection tools, particularly affecting English language learners, but Turnitin claims to have trained its detector on diverse writing samples.

Some universities have paused using Turnitin’s AI detector due to concerns about false positives and biases. The company maintains it has a false positive rate of less than 1%.

Students also utilize AI chatbots for research purposes which cannot be considered plagiarism.

Overall
The situation shows the challenges faced by educators in detecting and addressing AI use in submissions.

There’s a need for nuanced discussions between educators and students about the ethical use of generative AI in academic writing, rather than relying solely on detection tools.