From Ethics to Agency: Participatory Design of a Teacher Training Course for AI in Education

Abstract

Artificial Intelligence is increasingly being integrated into educational settings, yet its ethical implications and impact on pedagogical agency remain underexplored. This thesis investigates the ethical challenges and agency-related concerns in education through an educational design research process, with the aim of developing a teacher training course for K-12 educators, designed through their own voices. The study begins with a systematic literature review (2011–2022), conducted using PRISMA guidelines, which maps the current state of research on AI in education. This phase identifies substantial gaps in ethical frameworks, teacher-specific guidance, and the preservation of educational agency. Building on this foundation, the research adopts a participatory futures methodology, using the Delphi Method to co-construct eight future scenarios. These scenarios explore the socio-technical imaginaries shaping AI's pedagogical implications, including issues of equity, assessment, student voice, and professional autonomy. Subsequent research phases engage teacher educators through iterative focus groups, exploring how AI alters agency dynamics, subjective, intersubjective, and collective, within educational contexts. Findings reveal a pressing need to move beyond dominant techno-solutionist narratives and instead support teachers in reclaiming their roles as ethical and relational agents. These insights inform the co-design of a professional development course, which integrates dialogic, experiential, and reflective learning practices. The course is hosted on a custom-designed Canva platform and structured around a three-layered framework of educational agency, offering educators conceptual and practical tools to critically engage with AI. By foregrounding the symbolic, relational, and ethical dimensions of education, this thesis argues that responsible AI integration must not only be technically sound but also aligned with the core purposes of education: subjectification, qualification, and socialisation. It proposes that sustaining teacher agency requires special attention to the preservation and care of the educational lexicon, one that sustains complexity, openness, ethical discernment, as well as desire and memory in the face of algorithmic pressures. For it is through desire that alternative imaginaries of socio-technical systems and comprehensive educational ecosystems are made possible. This dissertation contributes four main outcomes: (1) a comprehensive ethical mapping of AI in education, (2) a participatory ethical dilemma toolkit, (3) a conceptual framework of agency in AI-mediated education, and (4) a context-responsive, agency-centred professional 8 Ana Mouta. 2025 development course for K–12 educators. Together, these outcomes constitute a theoretically grounded and empirically informed contribution to ongoing scholarly and professional efforts aimed at cultivating educational environments in which decisions regarding the use of AI, and the conditions under which it is integrated, are co-constructed through dialogic, participatory processes that uphold educational purpose, human agency, and the democratic ethos of schooling. It counters the depoliticising and deprofessionalising tendencies of technocratic models by supporting teachers in critically engaging with AI, resisting unreflective automation, and challenging algorithmic normativisation.