Teacher AI tool linked to lower student motivation in Turkey trial

A randomized school trial in Turkey found that giving teachers access to a generative AI support tool reduced students’ intrinsic motivation and did not improve average academic performance. Conducted during the spring 2025 semester, the study covered 193 teachers, 2,816 students and 14,198 student-course observations across 14 schools, with teachers assigned to business-as-usual teaching, AI access, or AI access plus weekly reminders and usage feedback.

The GPT-4o-powered tool included Turkish Ministry of Education curriculum materials and could support lesson planning, assessments, feedback, differentiated instruction and administrative communication. Students in AI-assisted courses rated classes as less important, less enjoyable and less interesting, with intrinsic motivation down 0.11 standard deviations. Average performance did not change significantly, but students taught by lower-performing teachers scored 0.129 standard deviations worse and also reported lower confidence.

Usage logs suggested teachers mostly used the system to generate teaching materials rather than deepen instruction. The researchers said many interactions were short and focused on outputs such as lecture materials, homework, exams, syllabus design and student reports, raising concerns that generic AI content may substitute for pedagogical effort. The findings point to the need for stronger training, guardrails and designs that reinforce teacher judgment rather than simply speeding up content production.