Abstract:

The rapid expansion of generative artificial intelligence in higher education has increased the need to move beyond ad hoc experimentation toward pedagogically grounded and institutionally supported integration models. Therefore, it is important to examine cross-institutional patterns of AI use in order to understand how these practices are shaped across different higher education contexts and how they relate to teaching design and student engagement.

This study examines how AI is used in teaching across three Estonian higher education institutions and to what extent these practices support students’ cognitive engagement. The analysis is based on a secondary comparative analysis of previously published institutional studies, combining quantitative survey data and qualitative lecturer reflections. The ICAP model is used as an interpretative framework to classify reported teaching practices according to levels of cognitive engagement.

The results show that AI is used predominantly for lecturer preparation, including material development, lesson planning, and assessment design. Student-facing uses are present but mainly limited to individual tasks. Within the ICAP framework, most reported practices correspond to passive or active levels of engagement. Constructive uses occur when tasks require students to revise, critique, or extend AI-generated content, but these practices are not systematically embedded. Interactive uses, such as collaborative evaluation of AI outputs, are rare. Across institutions, concerns related to academic integrity, assessment reliability, and unclear boundaries of acceptable AI use shape lecturers’ pedagogical decisions. Prior AI-related training is associated with greater confidence and lower perceived risk, yet training alone does not eliminate assessment-related concerns. Institutional governance and policy clarity influence whether AI use remains restricted or is integrated into teaching design.

The findings indicate that the educational value of generative AI depends primarily on instructional design and institutional conditions rather than on technological availability. Without intentional task design and structured guidance, AI is likely to function mainly as a lecturer support tool rather than as a means of promoting higher levels of cognitive engagement.

Keywords:

ICAP framework, institutional governance, instructional design.