Project R-14463

R-EduAI - Socio-technological reference framework for advancing responsible AI in education (Research)

Educational institutions and developers in the field of Artificial Intelligence in Education (AIED) are increasingly faced with the imperative of creating 'responsible' AIED systems. One way to fulfill this responsibility is by enhancing the 'intelligibility' of these systems. However, educational institutions often lack the necessary governance capabilities required for shouldering this responsibility. Conversely, developers find it challenging to design AIED systems that are truly intelligible because they struggle to assess the impact of these systems on individuals, educational institutions, and society as a whole. Furthermore, developers tend to lean towards creating generic AI systems rather than tailoring them to the specific objectives of educational institutions, which further complicates the development of intelligible AIED systems. Our overarching research objective is safeguarding and strengthening public interest values in Artificial Intelligence in Education (AIED). Our central research question is: How can we make Ethical, Legal, and Social Aspects (ELSA) actionable in the context of AIED? We intend to address this question by constructing a socio-technological reference framework that emphasizes the development, integration, and assessment of intelligible AIED systems. Our ultimate goal is to bolster the capacity of educational institutions to govern AIED responsibly and to empower developers to design responsible AIED systems. To attain this objective, we place a strong emphasis on 'mediation', and bridging the gap between end-users and innovation. We will conduct thorough desk research and engage in expert interviews to delve into legal and normative aspects, the underlying political-economic landscape, the various stakeholders involved, and technological affordances. Through participatory workshops, we aim to uncover the challenges related to ELSA, as well as the awareness, attitudes, literacy, and practices of educators and developers. Finally, we will analyze 'boundary objects' through qualitative semi-structured expert interviews as a means of bringing clarity and insight to this complex domain.

01 January 2024 - 31 December 2025