Living catalogue on emerging AI Practices in Education
The living catalogue is a practical guide to research-based emerging AI practices in education, curated and regularly updated by the Center for Learning Sciences at EPFL.
The use of generative AI in education is emerging fast, but not all uses are created equal. Some applications replicate existing tasks, while others fundamentally reshape how we teach and learn. The SAMR framework (Substitution, Augmentation, Modification, Redefinition; Puentedura, 2009; Belkina et al., 2025) offers a useful way to understand this spectrum.
At the substitution level, generative AI can replicate traditional tasks without altering their nature. For example, using ChatGPT to generate quiz questions instead of writing them yourself. Augmentation brings improvements to the efficiency or quality of processes. An example is the systematic alignment of feedback with pedagogical principles. Modification involves significant alteration of the process, like using generative AI to provide students with interactive feedback or automated assessment. And at the highest level, redefinition, AI enables entirely new kinds of educational experiences that were previously impossible, such as educational simulations.
To help both researchers and educators make sense of this rapidly evolving landscape, we’re building a living catalogue of studies on the use of generative AI in education. It spans the full range of SAMR levels, from basic substitution (e.g., evaluating the quality of AI-generated assessments) to fully transformative applications.
More than just a database, the living catalogue is a practical guide to research-based emerging AI practices in education, curated and regularly updated by the Center for Learning Sciences at EPFL.
- Expert-defined practices: We identify key trends in how generative AI is being used in educational settings, grounded in current research and theory.
- AI-assisted synthesis: We use large language models to process and extract relevant information from a wide body of academic literature.
- Real-world applications: Browse and filter examples of how generative AI is being implemented across institutions, disciplines, and contexts.
- Key outcomes: Learn what works, what doesn’t, and what to watch out for—effectiveness, accuracy, limitations, and implementation recommendations.
- Systematically updated: To adapt to a rapidly evolving field, we automatized our research pipeline all the way from database queries to structured data extraction.
