Article Text
As tools such as ChatGPT become more common at McLean High School, teachers face a difficult question: How can they identify improper AI use without falsely accusing students?
Fairfax County Public Schools permits Turnitin, a plagiarism-checking platform that includes an AI-detection feature. However, the district does not support using AI detectors as proof of misconduct. Despite that guidance, students are facing more accusations based on software-generated indicators rather than a teacher’s professional judgment.
AI detectors analyze patterns in sentences and the overall structure of a piece of writing. They do not understand a student’s ideas or reliably recognize an individual voice. Turnitin itself cautions that its AI-detection results do not definitively prove that AI was used.
That uncertainty can produce serious consequences. A false cheating accusation can cause stress and damage trust between students and teachers. The risk may be particularly pronounced in writing-intensive subjects such as English and foreign languages, where students’ styles can naturally change and improve.
Teachers are often better positioned to notice suspicious changes because they become familiar with each student’s work.
“Teachers can trust their professional judgment because they become familiar with each student’s own voice. When they notice something out of the ordinary, that is more reliable than an AI detector,” school-based technology specialist Bridget Donoghue said.
Some teachers turn to detection software in hopes of grading fairly and preventing students who use AI from gaining an advantage. But inconsistent results can also flag students who completed their assignments independently.
“Sometimes, I feel like teachers [who use AI detection tools] are just guessing if my work is AI or not. It makes me nervous, especially on big essays,” an anonymous junior said.
The technology also struggles to keep pace with rapidly improving AI systems. A detector that appears effective today may become less useful as AI-generated writing grows more humanlike. As a result, the software may miss improper AI use while identifying authentic student work as suspicious.
“AI is just becoming so much smarter, and it’s evolving every day. It’s impossible for those detectors to be 100% accurate. We can’t rely on those,” Donoghue said.
Unusual writing styles can create another problem. An essay with advanced vocabulary or a highly organized structure could be incorrectly labeled as AI-generated simply because it differs from what the software expects.
The emphasis on detection can also distract from learning. When a software score receives more attention than a student’s reasoning, meaningful feedback and academic growth can become secondary.
“I was once accused of using AI for an assignment just because a random website said it was AI-generated. It felt unfair, and the whole situation was really stressful,” the anonymous junior said.
FCPS does not approve AI detectors for disciplinary use because of reliability and student-privacy concerns. Although teachers may use Turnitin to check for plagiarism, its AI-detection feature is not intended to serve as evidence of cheating.
“Fairfax County doesn’t allow us to use AI detectors because of student privacy issues and unreliability,” Donoghue said.
Teachers have other ways to evaluate whether students produced their own work. They can observe the writing process, assign work during class and become familiar with how each student develops ideas. Those approaches can provide more context while reducing stress and preserving trust.
AI-detection software should not determine whether a student cheated. Careful teacher judgment and assignments that reveal students’ thinking offer a fairer response than relying on an uncertain computer-generated score.