Artificial intelligence is often touted as a way to make teachers more effective by helping them write lesson plans, create teaching materials, and provide feedback to students in seconds. But one of the first randomized studies to test AI in real classrooms found that it can also interfere with learning. Students whose teachers were given access to an AI teaching assistant felt affected less motivated to learn.
The damage was particularly pronounced among students whose teachers were already weaker teachers, based on their performance before the experiment began. The researchers found that their students also scored lower on standardized final exams.
“Teachers, like students or programmers, may be using AI as a crutch,” said Alp Sungu, lead author of the study and assistant professor at the Wharton School at the University of Pennsylvania. “Instead of doing the actual work, they use AI to delegate the task, and that reduces the quality of their teaching.”
A draft of the study, titled “Generative AI Can Harm Teaching,” was published online in June and has not yet been published in a peer-reviewed journal. It reflects Sungu’s much-discussed 2024 research on how students’ use of AI affects learning.
“Students are using AI as an answer machine, not a learning tool, and therefore it is detrimental to learning,” Sungu said. “Here I think teachers may be using AI as a material generation machine for homework, lecture notes, lesson plans and syllabus. Instead of improving their own output, they are using AI as a substitute with very minimal interaction and hence the quality of the output is not good enough.”
Related: Faster Solutions, Lower Test Scores: How AI Is Undermining Math Skills
Sungu’s experiment, conducted with fellow University of Pennsylvania researchers including educational psychologist Angela Duckworth, followed 193 teachers and more than 2,800 middle and high school students at a private school chain in Turkey in spring 2025.
Teachers were randomly assigned to receive access to a ChatGPT-based teaching assistant tailored to Turkey’s national curriculum or to continue teaching as usual. Over a period of 10 weeks, teachers used the tool primarily to create lecture notes, homework and exams.
Students whose teachers had access to the AI tool rated their lessons as less enjoyable, less interesting and less important than students in the control group. The decline in intrinsic motivation was moderate, but greater for students of teachers who were already heavy AI users before the experiment began.
Overall, the average academic performance did not change. But for teachers whose students had lower grades before the experiment – an indicator of lower-performing teachers – both student performance and self-confidence declined. Academic achievement was measured by externally administered standardized examinations, eliminating the possibility that these teachers had different assessment standards.
The study cannot explain exactly why the quality of teaching deteriorated. The researchers did not observe classrooms or analyze the AI-generated materials used by teachers. But Sungu suspects teachers may have given up one of their most effective tools.
“When you start using AI-generated material, you lose your personal voice,” Sungu said. “Maybe it’s technically good enough, but it doesn’t really carry your own style. If everything is very uniform, it just becomes a little boring.”
A possible explanation for the weaker academic performance of students with low-performing teachers, Sungu said, is that stronger teachers view the AI results as a first draft, revise it and adapt it to their classrooms. He suspects that weaker teachers may be more likely to use AI-generated material as is.
Related: AI praises black students more and criticizes them less
This study is not a clean comparison between teaching with and without AI. Teachers in the control group were free to use other AI tools. This enabled a comparison between access to a tailored AI assistant and what teachers wanted to do themselves. If anything, Sungu said, these results may underestimate the risks of teachers relying heavily on AI-generated materials.
Still, Sungu warns that it would be a mistake to conclude that “AI is terrible and will ruin education.” He sees another lesson: access to AI technology alone does not improve teaching.
The challenge is to help teachers use AI in a way that preserves human judgment and creativity. This requires teacher training programs, guidelines and better interfaces.
“Right now there is concern about how teachers are using it organically,” he said.
Sungu says he personally uses AI in his university teaching to create interactive games and surveys that would otherwise take too long to create. “When I first get the copy, it looks great,” he said. “And then if I don’t engage with it, the examples and the numbers don’t make sense. I end up spending just as much time improving the output or adapting it to my class.”
“It’s not a time saver,” he said.
Contact the staff Author Jill Barshay at 212-678-3595, jillbarshay.35 on Signal or barshay@hechingerreport.org.
This story about AI in education was produced by The Hechinger Reporta nonprofit, independent news organization covering education. Sign in Proof points and other Hechinger newsletters.
This article first appeared in the Hechinger Report and is republished here under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
https://hechingerreport.org/proof-points-ai-in-teaching/
