From AI-Generated Student Analytics to Teacher Trust: Evidence from a Longitudinal Classroom Case Study

This case study examined how teacher trust in student analytics could be built when using Artificial Intelligence (AI) generated assessment. The case study included two data sources: (i) teacher evaluations of AI-generated student reports and (ii) a semi-structured interview with a science teacher who used the system in their classroom for three semesters (first pilot semester and two semesters of implementation). The teacher demonstrated an agreement rate exceeding 80% with Generative Artificial Intelligence (GenAI) generated reports. However, the teacher also stated that she expected the reports to not only describe student characteristics but also provide actionable pedagogical recommendations to support the teaching process. In conclusion, the study provided evidence that the teacher trust in GenAI based student analytics could be developed as the teacher adopts AI based practices in her classroom. When GenAI tools are implemented for a longer period, they have potential to build trust underlined by other studies. Including GenAI represented students' learning characteristics and replicated it consistently across different students. The findings indicate that GenAI-supported learning analytics systems to be developed in the future should be designed not only as tools to describe student behavior, but also as explainable and pedagogical decision support systems that support teachers' professional judgment.