The Future of Education

Artificial intelligence is increasingly influencing higher education, requiring instructional designs that integrate AI-supported learning environments [1]. This paper presents an experience-based account of how notebook-based large language models (LLMs) [2] can be systematically employed to create structured and transparent learning pathways in a graduate-level UX Design course for eHealth. The approach uses clearly defined, fine-grained learning objectives, each formulated as explicit guiding questions, enabling students to understand precisely the competencies to be acquired and the corresponding assessment criteria. During self-study, students work with curated LM notebooks that provide explanations, contextual examples, and micro-tasks aligned with each learning objective. Classroom sessions focus on the deliberate practice of these competencies, with the instructor taking on an active supervisory and corrective role. This structure addresses a central challenge of AI-supported learning: ensuring authentic student competence rather than reliance on automated solution generation, thereby mitigating academic integrity risks [3]. Assessment occurs exclusively under controlled conditions. Students perform usability and accessibility evaluations during supervised sessions, and each student completes an individual oral examination. These oral exams serve as a verification mechanism ensuring that learners have acquired genuine conceptual and procedural understanding, reflecting current discussions on the renewed relevance of oral examinations in the age of AI [4]. This combination prevents academic misconduct and reinforces assessment validity in AI-enhanced educational settings [3], [4]. The course model demonstrates that combining LM-supported preparatory work with supervised practical assessment and oral examination can enhance learning outcomes, maintain academic integrity, and provide a robust pedagogical framework for integrating AI into higher education.