Abstract:

Generative artificial intelligence is rapidly transforming higher education. However, most institutions are being pushed toward commercial, proprietary, and opaque AI systems that offer little control over data flows, model behaviour, or pedagogical alignment. To ensure that European legal and ethical frameworks are respected, a different approach is required in order to protect autonomy, privacy, and long-term sustainability.

At Karlstad University’s Centre for Teaching and Learning hosted in the library, we have developed a digital-sovereign AI chatbot grounded in open standards, open-source technologies, and institutionally governed infrastructure. The project is not only a technical experiment but also a strategic intervention: an attempt to create an AI service that strengthens—rather than compromises—the core public values of higher education.

Our prototype integrates Learning Tools Interoperability (LTI) for secure embedding into the learning management system and uses WordPress as a modular, extensible knowledge layer. The system is designed so that universities—not third-party vendors—own the data pipeline, determine governance rules, and shape the educational use cases.

Another central innovation is the structured ingestion of Open Educational Resources (OERs). By feeding the AI model with high-quality, openly licensed course materials, information-literacy content, and student-support resources, we can dramatically improve response accuracy and reduce hallucinations. This approach creates a pedagogically aligned assistant that is grounded in the institution’s own knowledge ecosystem, yielding more trustworthy and context-relevant outputs.

The presentation will highlight our architectural choices, implementation experience, user feedback, and governance considerations, with a focus on the wider communities at the crossroads Technology, Education and Development. Specifically, we will address what it means for a university to “own” its AI infrastructure, how LTI and open-source platforms ensure long-term sustainability and how structured content ingestion reduces risks, improves learning outcomes, and empowers educators as well as learners.

This talk argues that European universities can and should build AI systems aligned with public-sector values, and that the IATED community is uniquely positioned to drive such an ecosystem. Rather than relying solely on commercial closed models, we can collectively create transparent, sustainable AI services that reflect our shared mission to support research, education, and knowledge as public goods.

Keywords:

Customized Chatbots, Open Educational Resources, Higher Education, Innovative Pedagogy.