by Patricia Santos, University Pompeu Fabra, Spain.
The rapid rise of Generative Artificial Intelligence (GenAI) tools is fundamentally reshaping how students learn, research and collaborate. While these tools offer powerful new possibilities, they also pose significant challenges to academic integrity, particularly at higher levels of education where originality, critical thinking and ethical discernment are paramount. Amid these growing tensions, a team of researchers and educators from the School of Engineering at Universitat Pompeu Fabra (UPF), members of the Interactive and Distributed Technologies for Education (TIDE), have published a pilot study exploring a promising solution: Learning Agreements (LAs) for responsible GenAI use.
This pilot study was recently published as a peer-reviewed article in Innovations in Education and Teaching International: Beardsley, M., Santos, P., Amarasinghe, I., Theophilou, E., Vujovic, M., and Hernández-Leo, D. (2025). A Learning Agreement for generative AI use in university courses: A pilot study. Innovations in Education and Teaching International, 1–20. The publication illustrates the design, implementation and evaluation of Learning Agreements (LAs) as a pedagogical strategy for guiding responsible GenAI use in a first-year university course. The initiative has been led and developed within TIDE and, thanks to the dual role of Patricia Santos (as both a researcher in TIDE and Head of the School of Engineering’s Innovation and Teaching Quality Unit) there are concrete plans to scale the approach across other courses and programmes within the school.
Patricia Santos explains: “At the core of this project is the principle of co-design. It involves engaging students, instructors and programme coordinators in a collaborative process to define clear expectations and boundaries for the use of AI in academic work. This participatory approach takes into account the ethical and disciplinary complexities of each context, empowering learners to critically reflect on how GenAI tools can enhance, but not replace, their academic growth.” Dr Santos adds, “Imposing fixed rules is not enough. What we need are co-created, flexible frameworks that respect disciplinary diversity, support student agency and foster ongoing ethical reflection.”
The study, conducted during the 2023–24 academic year, invited engineering students to complete a structured Learning Agreement (LA) at the start of the course, outlining clear expectations around GenAI use, citation and ethical considerations. The approach was embedded into coursework and assessed through surveys, group projects and qualitative feedback. The findings were insightful: students showed a significant increase in GenAI awareness and usage over the term, and most appreciated the opportunity to explore AI in a structured and critical manner. However, compliance with the agreement’s ethical provisions was mixed: while students declared whether they had used GenAI tools, only a small minority followed through with proper citation or reflected on the limitations of the tools used.
These gaps reflect a broader challenge facing higher education: although general guidelines on academic integrity exist, few offer concrete, practical strategies tailored to the use of AI in specific course contexts. In response, the UPF team is extending its work beyond the initial pilot to other undergraduate courses and to the doctoral level. They are planning to develop LAs designed to support both undergraduate and PhD students, as well as their teachers and supervisors.
This initiative aligns with emerging global research and policy frameworks, such as the EU AI Act, which call for more nuanced, context-aware approaches to AI governance in education. By focusing on autonomy, competence and relatedness, the LA model supports not only ethical compliance but also student motivation and engagement.
Led by Dr Santos and supported by the TIDE research group, the next phase will involve the co-creation of a reusable toolkit, including templates, case studies and training materials that can be adapted across disciplines and institutions. These tools are intended to foster inclusive, ethical learning environments where AI can be explored critically and creatively.
In short, Learning Agreements (LAs) for GenAI use offer a promising, student-centred strategy for addressing one of the most urgent pedagogical questions of our time. They move beyond static policies and towards a culture of shared responsibility, transparency and ethical dialogue. As AI continues to transform academic practices, these agreements may well become essential instruments for ensuring that integrity evolves alongside innovation.
To learn more about the project or collaborate on co-designing Learning Agreements (LAs) for Higher Education, contact Dr. Patricia Santos at patricia.santos@upf.edu.
Patricia Santos is a Tenure Track professor and Ramón y Cajal fellow in the TIDE group within the Department of Engineering at the University Pompeu Fabra. Her research includes technology-enhanced learning, human-centred learning technologies, learning design and the use of learning analytics to support education.
