Schools are rushing to adopt artificial intelligence to save teachers time, but a classroom trial suggests that relying too heavily on AI backfires. When teachers use AI as a shortcut to generate assignments and lesson plans, students lose motivation and, in some cases, score lower on exams.
What Happened
A randomized trial followed 193 teachers and over 2,800 middle and high school students over a ten-week period in Turkey during the spring of 2025. According to the working paper, Generative AI Can Harm Teaching, half of the teachers had access to a custom ChatGPT-based teaching assistant, while the other half taught as usual.
The researchers, led by Wharton School assistant professor Alp Sungu and educational psychologist Angela Duckworth, discovered that students whose teachers used the AI tool rated their classes as less enjoyable, less interesting, and less important. For teachers who were already weaker instructors, their students' confidence and standardized exam scores both declined. Sungu warned that teachers may be using AI as a crutch to generate homework, lecture notes, and syllabus materials with minimal personal effort, which lowers the quality of the curriculum. This study builds on Sungu's prior research showing that students' use of AI is harming learning when they treat tools as simple answer machines.
The Bigger Picture
Using AI as a substitute for prep work damages the classroom experience, but research shows this technology can succeed when used as a collaborative partner rather than an automated replacement. For example, high school English teacher Eric V. uses Google Gemini to escape writer's block and personalize lessons, while third-grade teacher Monica H. utilizes targeted prompts to adjust reading levels and customize unit plans. EdTech experts emphasize that AI should only serve as a tool for "fast first drafts" that require human revision.
Academic studies back up the importance of proper teacher preparation. Research on preservice teachers found they could successfully refine AI-generated lesson plans through collaborative reflection, rather than using unedited outputs. Another study highlights that formal teacher training improves AI literacy and digital material design competencies, ensuring educators have the skills to tailor these resources. Professional development is shifting teacher training from simple content creation to developing authentic, process-oriented assessments.
When implemented thoughtfully, AI does not have to hurt motivation. A systematic review found that AI-supported digital learning environments can significantly improve science learning outcomes, though the results depend on the school context. Research shows that real-time AI feedback can actually boost student motivation and confidence when paired with interactive learning, while a student's own technological literacy plays a massive role in whether AI tools succeed or fail.
What This Means for Families
This research shows that AI is not a magic wand for education. When teachers automate their relationship with students or dump unedited AI materials into the classroom, the human connection that drives learning disappears.
As we previously reported on teacher-made learning tools, the most effective educational technology is built with a deep understanding of student needs. Parents and school administrators must look closely at how local classrooms are integrating these systems. If AI is used to free up teachers for more one-on-one time with kids, it is a win. If it is used to avoid doing the hard work of teaching, student performance will suffer.
What You Can Do
Parents should ask teachers if and how they use AI to generate assignments, lesson plans, or student feedback.
Educators using AI should treat outputs as rough drafts, using frameworks like the PARTS framework to refine content.
School administrators must prioritize professional development that focuses on collaborative lesson planning and robust rubric design, rather than simple text-generation shortcuts.