Ask any teacher what’s changed most in the last two years, and “AI” comes up almost immediately — usually attached to a worried story about a suspiciously polished essay. But new research suggests the panic over cheating may be pointing at the wrong problem.
Education researcher Brett DeJager surveyed 303 educators in Wisconsin and 132 more nationally between spring 2025 and spring 2026, and the results, published via The Conversation and picked up widely including by Phys.org, paint a more complicated picture than a simple cheating epidemic.
What Teachers Are Actually Worried About
Plagiarism concerns were widespread — 65% of Wisconsin respondents and 74% of the national sample said they worried about it. But dig one layer deeper and a different concern shows up almost as often: 47% of Wisconsin teachers and 53% nationally flagged something more specific — “difficulty in assessing student learning when AI is used.”
In other words, the issue isn’t just that some students are using AI to cut corners. It’s that even honest, AI-assisted work makes it harder for a teacher to tell whether a student actually understands the material. Between 29% and 40% of respondents noted increased student reliance on AI tools, and 19% to 33% observed reduced critical thinking or problem-solving in their classrooms.
Meanwhile, policy hasn’t caught up: only 33% of Wisconsin districts and 29% nationally had formal AI policies in place at the time of the survey.
The “Deeper Issue”
DeJager’s central argument is that finished work — a clean essay, a solved problem set — has stopped reliably indicating what a student actually knows. Polished output used to be a decent proxy for understanding. It no longer is.
His proposed fix isn’t a ban or a detection arms race. It’s redesigning how learning gets measured in the first place: assignments that require students to demonstrate independent writing in real time, explain their reasoning out loud, complete meaningful work in class, or document their process — not just hand in a final product and trust that it reflects their own thinking.
What This Means for Parents and Students
For families, the takeaway isn’t that AI use is automatically dishonest — plenty of legitimate uses exist, from brainstorming to checking grammar. The more useful question to ask a student isn’t “did you use AI,” but “can you explain this without it.” Schools that lean into that kind of verification — oral checks, in-class writing, process documentation — are likely to be ahead of the curve as AI policy continues to lag behind AI adoption.
Sources: The Conversation / Phys.org — Teachers are worried about students cheating with AI, but survey suggests the deeper issue is learning; survey by Brett DeJager, spring 2025–spring 2026.