AI in Higher Education Cookbook
This university-wide USO project examined how generative AI (GenAI) technologies are reshaping teaching, learning, and assessment in higher education. The project focused on building reflective AI awareness among both staff and students. This entailed not only understanding what GenAI tools can do, but also developing the critical awareness required to decide when, how, and whether their use is educationally appropriate.
Project overview
Over a period of two and a half years, the project brought together researchers, teacher, and students from multiple faculties to design and run course-based educational interventions. These interventions explored the opportunities and risks of GenAI in real educational settings, generating empirical insights into AI-supported teaching practices, student learning, assessment design, and AI literacy. In doing so, the project also helped establish a cross-faculty network of educators and researchers that continue through the Special Interest Group (SIG) GenAI in Higher Education.
Rather than aiming for universal rules or tool-specific prescriptions, the project prioritised practice-based knowledge: what happens when teachers and students actually work with GenAI in real-world contexts.
Educational cookbook
The work carried out during the project culminated in an educational cookbook. This cookbook collects and presents interventions developed and tested across the project as ‘recipes’ with structured descriptions of pilots that document their education context, design choices, outcomes, and lessons learned.
Each recipe captures:
- the educational challenge or opportunity that prompted the intervention,
- how GenAI was incorporated into teaching or learning activities,
- what happened in practice, and
- what others can learn from the experience.
The cookbook does not promote best practices in a narrow or prescriptive sense. Instead, it makes visible the design reasoning, trade-offs, and constraints educators encountered when experimenting with GenAI. In doing so, it supports reflection and adaptation rather than replication.
The cookbook is organized into four thematic chapters, corresponding to the project’s work packages. It is intended for teachers, students, programme directors, and educational developers who want to move beyond abstract debate and engage with practical examples of GenAI in practice. Visitors can browse each theme, explore how colleagues experimented with AI, learn from both successes and failures, and use these insights as inspiration for their own educational contexts.
Should you want to build on these examples in your own practice, the following reflection questions may be helpful if you are considering trying GenAI in your teaching. They are based on colleagues’ experiences shared in this cookbook.
- Students’ use: What do I actually know about how my students use GenAI, what they want from it, and which (ethical) concerns they have?
- Learning mechanism: Which part of the learning process does GenAI support or take over (e.g., idea generation, source selection, writing support, debugging), and what learning mechanism does that affect?
- Tool fit and limits: Do I understand what this tool can and cannot do well in my context, and how I will check the quality of its output?
- Teachers’ use: If I introduce GenAI in my course, is it likely to assist me as a teacher and reduce workload, or mainly shift it? Does it change my role as a teacher, or how students view my role as a teacher?
- With vs. without AI: How does this fit the learning trajectory? At the start, you may want students to practice core skills without GenAI; later, GenAI can be appropriate support once the skill is established.
- Assessment and alignment at course and programme level: If GenAI is introduced in my course, how will this affect the validity and fairness of my assessment of the intended learning outcomes? Do learning activities, goals, and assessment remain aligned at the course and programme level?
The Cookbook
Coordinator: Sjoerd Dirksen
This theme brought together interventions that explored how GenAI was used to support teachers in designing courses, preparing learning activities, and facilitating instruction. The pilots examined AI-assisted content generation, instructional planning, and simulated student interaction.
Across the interventions, a recurring finding was that GenAI functioned best as a complementary and reflective tool, rather than as a replacement for pedagogical judgement. Effective designs provided clear guidance on responsible use, embedded AI used within well-structured tasks and required teachers and students to critically evaluate AI outputs. These recipes illustrate how AI can reduce certain workloads while simultaneously strengthening instructional quality and AI literacy.
Recipe: Teaching and learning collection | Using Generative AI for Legal Tutorial Assignments
Recipe: Teaching and learning collection | AI index, bias, duurzaamheid van AI
Coordinator: Sergey Sosnovsky
This theme focused on how GenAI was used by students as a learning support tool. Interventions explored AI-generated feedback, writing support, motivation, and student perceptions across multiple disciplines.
The pilots demonstrated that while GenAI could support information access and task efficiency, it did not automatically lead to deeper learning. Learning benefits depended heavily on teacher-designed scaffolding, explicit prompting strategies, and structures reflection. Without such guidance, AI use risked bypassing productive struggle and undermined learning goals. The recipes in this chapter highlight both the possibilities and the limits of AI-supported learning.
Recipe: Teaching and learning collection | Generative AI and Basic Psychological Needs of BA students
Coordinator: Laurence Frank
This theme examines how GenAI affected the validity, reliability, and fairness of assessment. Interventions addressed assessment at three levels: individual courses, degree programmes, and university-wide policy context.
The pilots showed that GenAI posed serious challenges to traditional assessment formats, particularly where final products could be generated without demonstrating underlying competencies. At the same time, they demonstrated that carefully redesigned processes, such as staged assignments, supervised components, and reflective use of AI could preserve validity while supporting AI literacy. These recipes provide practical models for AI-aware assessment design.
Recipe: Teaching and learning collection | Generative AI and assessment of AE
Coordinator: Karin van Es
This theme addressed AI literacy as a core academic competence rather than a technical skill. Interventions focused on tool criticism, ethical reflection, and disciplinary context, drawing on both empirical research and classroom experimentation.
Work in this theme included university-wide surveys, qualitative studies, and educational pilots that explored how students and teachers understood, trusted, and critiqued GenAI. The resulting recipes emphasise that responsible AI use requires more than instructions, it depends on explicit discussions of values between educators and students, and co-created guidelines within educational communities. We see how the introduction of GenAI into our education highlights the shortcomings of the current system, presenting an opportunity to rethink educational practices and goals. More importantly, it stresses the need for Utrecht University to make available open source GenAI models that protect data sovereignty and privacy.
Recipe: Teaching and learning collection | Generative AI and critical coding
Recipe: Teaching and learning collection | Generative AI and cultural biases
Recipe: Teaching and learning collection | Generative AI en verantwoord chatbotgebruik in hoger onderwijs
Recipe: Teaching and learning collection | Generative AI being used for essay co-writing
Recipe: Teaching and learning collection | Reflectietool voor generatieve AI in het onderwijs
Additional products
Teaching and learning collection | Towards unbiased assessment of adaptive expertise
GSLS GenAI Tutorials for Students
GSLS GenAI Tutorials for Teachers
Would you like to keep the conversation about GenAI in Edcuation going? Join the SIG GenAI!
References
Kasneci, Enkelejda, et al. “ChatGPT for good? On opportunities and challenges of large language models for education.” Learning and Individual Differences 103 (2023): 102274.
Van Es, Karin, Mirko Tobias Schäfer and Maranke Wieringa. 2021. “Tool Criticism and the Computational Turn: A ‘Methodological Moment’ in Media and Communication Studies.” M&K Medien & Kommunikationswissenschaft 69 (1): 46-
Van Es, Karen and Dennis Nguyen, ‘Setting an agenda for Generative AI in education’ DUB, 04/12/2024
Central AI policy Central AI policy
All AI-related activities on this page must be implemented in line with Utrecht University’s central AI policy and ethical code.
Responsibility for appropriate tool choice, data protection, transparency, and assessment use remains with the instructor.
You are free to share and adapt, if you give appropriate credit and use it non-commercially. More on