Mind the Trust Gap: Identifying (Mis)alignments in Teacher-Student Views Toward Control and Agency in K-12 Classroom AI

Synthesis: As AI technologies enter K-12 classrooms, understanding how different stakeholders perceive these tools is critical. This paper identifies a significant trust gap between Teaching and Student Experience perspectives on control and agency in classroom AI systems. Students generally desire greater autonomy and flexibility when interacting with AI tools, while teachers prioritize oversight, monitoring, and structured control to maintain pedagogical alignment. These misalignments have direct implications for AI Literacy programs that must address both stakeholder perspectives to build effective Student Experience in learning environments.

Key Findings

• The researchers conducted a speed-dating study using storyboards with 16 school students and 15 school teachers in Germany to investigate alignments and misalignments in views on student-AI decision-making control in K-12 classrooms.
• Through explicit pair-matching analysis, they found that students and teachers had misaligned views on several key topics, including how much they trust AI and the social and emotional aspects of student learning with AI.
• Students emphasized the importance of their own decision making and of human teachers as gatekeepers of the classroom, arguing that teachers have a "better connection" to students and understand them as whole persons, not just through information detected by systems.
• Teachers also reported not fully trusting AI's decision making, citing the risk of incorrect assessment of students (e.g., assigning tasks that do not match actual knowledge levels), and stressed the importance of being able to intervene and override AI's automatic assignments.
• Contrary to students' preference for human teachers, many teachers worried that students would trust AI more than teachers themselves — the mirror image of the students' stated view.

Study Design & Method

The speed-dating study paired participants with a series of storyboard scenarios depicting classroom AI systems, eliciting rapid, repeated judgments across design situations. The pair-matching analysis then explicitly compared teacher and student responses on shared topics, revealing where views aligned and where they diverged. The design foregrounds the classroom as a multi-stakeholder environment in which the same technology is perceived differently depending on role.

What this means for practice

• Instructors. Treat control over classroom AI as something to negotiate with students rather than a default to configure, because the 16 students and 15 teachers diverged on how much they trust AI and on the social and emotional aspects of learning with it.
• Instructors. Keep an override for every automatic assignment and show students that it exists, since teachers cited incorrect assessment of students — tasks not matching actual knowledge levels — as their central risk with AI decision making.
• Instructors. Discuss what the system infers and monitors rather than leaving it to the tool, since students argued teachers understand them as whole persons in ways detected information cannot capture.
• Learners. Ask which data a classroom AI system used before accepting its placement or pacing decisions, and raise disagreement with your teacher instead of silently working around it.
• Instructors. Address the social and emotional dimensions of AI-mediated learning explicitly, and build trust through ordinary teacher-student relationships outside AI use — the authors report those relationships shaped how both groups viewed the technology.

Limitations

• Storyboards directed participants' thinking toward the classroom situations depicted, and the study used intelligent tutoring systems as its example, so views on other AI tools such as generative AI could differ.
• Data were collected in Germany, where AI-based tools are not yet fully realized in schools; the authors expect views to shift once teachers and students experience AI systems in classrooms.
• The samples were small and mismatched: two participating teachers taught only students younger than 12 while the student sample was 12 and above, and participants were compensated, which may have influenced what they shared.
• Older students generally shared more thoughts than younger students, which the authors suspect masked details that only younger students had considered.

Citation

Tomohiro Nagashima, Lisa Siegrist, Niklas Scholz, Shintaro Sato, Martina Vincoli, Man Su (2026). Mind the Trust Gap: Identifying (Mis)alignments in Teacher-Student Views Toward Control and Agency in K-12 Classroom AI.