German study on K-12 AI trust: Students and teachers don't align on control
Speed-dating storyboards with 31 participants reveal misalignments on AI decision-making control and trust.
Researchers led by Tomohiro Nagashima (affiliated with German institutions) published a study on arXiv (July 2026) examining trust gaps between teachers and students regarding AI control in K-12 classrooms. They used a speed-dating study design with storyboards, pairing 16 students and 15 teachers to compare views on student-AI decision-making control. The explicit pair-matching analysis revealed systematic misalignments: teachers and students disagreed on how much authority AI should have over classroom decisions, the level of trust appropriate for AI systems, and the importance of social-emotional factors in AI-assisted learning.
Key findings also showed that the quality of teacher-student relationships outside of AI use significantly shaped how each group perceived AI's role. The authors argue that designing effective classroom AI must account for these perspective gaps, as preferences depend on dynamic relationships among stakeholders. The study is set to appear in Proceedings of the ACM on Human-Computer Interaction (CSCW 2026), underscoring the complexity of human-AI interaction in educational settings and offering strategies to bridge alignment gaps.
- 16 students and 15 teachers in Germany participated in a speed-dating storyboard study to compare views on AI decision-making control.
- Misalignments were found on trust in AI, social-emotional learning aspects, and the influence of teacher-student relationships outside AI use.
- The study will be published in ACM HCI proceedings (CSCW 2026) and suggests strategies to fill perspective gaps for better classroom AI design.
Why It Matters
Designing classroom AI that both teachers and students trust requires bridging these hidden alignment gaps to ensure effective adoption.