Challenges and opportunities of using artificial intelligence in human rights education: Reflections from higher education practice

Synthesis: Kasa-Mälksoo argues that the hard problem AI poses for human rights education is pedagogical rather than technological. Working from more than ten years teaching human rights in a specialized master's program at the University of Tartu, and from her own integration of generative AI into course design, assignments and assessment across 2023–25, she examines AI through the UN's tripartite framework of education about, through and for human rights, tested against the principles of participation, equality and inclusion. The direct evidence is two classroom episodes. Written work grew polished while class discussion lost the "uncertainty and intellectual struggle" of a genuine encounter with difficult scholarship, and critical views that students never voiced in class appeared in end-of-course written feedback — what she reads as dialogic retreat into AI-mediated channels. Her conclusion is not prohibition but direction: human rights education should shape how AI is used, teach AI governance as a professional responsibility, and resist uses that bypass dialogue.

Key Findings

1. The challenge is pedagogical, not technological. AI enters an educational relationship that already depends on dialogue, trust and participation, and the article argues the harder question is how those conditions survive integration rather than which tools are adopted.
2. Polished written work outran spoken engagement. In one course, students' written outputs engaged the required HRE scholarship impressively while class discussion lacked the hesitation a genuine encounter with unfamiliar ideas tends to produce — a gap between the product and the understanding.
3. Critical views surfaced only in writing. End-of-course written feedback contained carefully articulated criticism that had not appeared in class, which the author reads as a shift into AI-mediated and anonymous channels rather than a gain in honesty.
4. A session-design episode raised formation, not integrity. Students asked to build HRE learning activities from COMPASS resources submitted professionally structured designs that showed little engagement with those sources, a professional-formation problem for future practitioners.
5. The through dimension is where AI bites hardest. AI shapes the conditions for trust and dialogue, and since didactic teaching already dominates higher education, integration can reinforce content transmission instead of participatory learning.
6. The for dimension reframes AI governance as professional responsibility. Using the EU AI Act and the Council of Europe Framework Convention, the author asks students to treat regulation as a domain they will shape, not an academic topic.
7. The educator's role grows, not shrinks. Digital skills are necessary but insufficient; the educator mediates between what technology makes possible and the pedagogical conditions for dignity, participation and transformation.

What the study is, and how it was done

This is reflective inquiry, not an empirical study: a focused review of policy and scholarly literature combined with reflection-on-action over a completed teaching cycle, grounded in Dewey and Schön via Greenberger's account of reflective practice. The setting is a specialized master's program in legal education in Estonia; from 2023–25 the author used GenAI tools for course design, assignments and assessment, and from 2024–25 made the human rights implications of AI an explicit curriculum component. The paper presents itself as practitioner-informed scholarship in a thin field and states plainly that its insights come from one educator in one program.

About, through, for: how each dimension behaves under AI

The tripartite frame does the analytical work. In education about rights, AI is an informational tool — chatbots answer questions fluently and can support personalized learning — but the author names the temptation to copy answers unexamined, notes that outputs depend on training data that may be dated or unrepresentative, and flags hallucination as most dangerous exactly where learners cannot yet judge it. In education through rights, AI is not neutral: it changes the conditions under which trust and dialogue are sustained, and where students and teachers are both experimenting invisibly, honesty "can no longer be assumed; it must be consciously modelled." In education for rights, AI becomes subject matter: the regulatory instruments give students concrete legal context for contesting how systems are governed, which shifts the register from understanding a problem to acting on it.

Two episodes, and what they reveal

The first episode concerns session design: asked to build HRE activities from established methodological resources, some students submitted polished design documents with little evidence of engaging those sources. The concern is not misconduct but professional formation — if AI generates the session without the reflective process of designing it, the competencies a practitioner needs were never developed. The second concerns dialogue: students who wrote impressively were visibly less secure in oral presentation, and the most critical observations arrived in writing rather than in class. Both describe one pattern, the appearance of knowing about human rights without the disciplinary engagement HRE exists to build, and both were met by naming the gap with students rather than banning tools.

What this means for practice

• Instructors. Model your own AI use openly — honesty must be demonstrated, not assumed — and assess the reflective process of designing a session or assignment, since a polished artifact produced without that process is where professional formation fails.
• Curriculum designers. Add AI governance as a subject of critical inquiry in HRE courses, using the EU AI Act and the Council of Europe Framework Convention as the concrete legal context students will work with as practitioners.
• Institutions. Align AI adoption with the participation, equality and inclusion principles that underpin the right to education, and pair any integration mandate with meaningful professional development, since digital skills alone do not carry the pedagogical work.
• Researchers. The written-versus-spoken gap this article surfaced — polished output alongside insecure dialogue — is testable with designs it does not attempt.

Limitations

• One educator's reflective account from one program at one university: no student data, no comparison group, no measured outcome, and the author presents it as practitioner insight rather than evidence of effect.
• Policy and literature are reviewed selectively rather than systematically, so the regulatory framing leans on the EU AI Act and the Council of Europe Framework Convention.
• Both classroom episodes rest on the author's recollection as course instructor, with no independent record of student work, discussion quality or AI use.
• The context is a European legal-education program in a specific regulatory environment; the author claims the insights may travel, not that they generalize as findings.

Connected Concepts

• Critical Pedagogy — the tradition the article works in, treating education as a site of power and of contestation
• Human AI Collaboration — AI as a participant in the pedagogical relationship rather than a neutral instrument
• Generative AI — the tool class integrated into course design, assignments and assessment across 2023–25
• Metacognition — reflective practice and reflection-on-action as both method and pedagogical aim
• Ethics — the rights-based frame, including privacy, non-discrimination and the right to education
• Equity — participation, equality and inclusion as the principles against which adoption is judged
• AI Use and Disclosure Statements — the disclosure question the author reframes as a condition for trust rather than a rule
• AI Governance — AI governance taught as a professional responsibility for future practitioners
• Hallucination Risk — why fluent answers are most dangerous where learners cannot yet judge them
• Learner Agency — learners positioned to help shape how technology develops, not just to understand it
• Legal Education — the disciplinary setting, where technical competence and ethical judgment meet

Connected Articles

• The Reflective Triangle Model: AI as a Cognitive Mediator in Teachers' Professional Learning and Learning-Community Development — reflective practice as a structure for teacher engagement with AI
• Teachers as reflective regulators of cognition: Understanding cognitive offloading in AI-augmented practice — teachers as reflective regulators of their own cognitive work with AI
• Critical AI Tutors: Empower or Enslave? — the critical-pedagogy reading of AI tutoring
• Don't believe the hype. AI myths and the need for a critical approach in higher education — critical scrutiny of AI claims in higher education
• Epistemic Dependence in AI-Mediated Learning — the dependence risk the polished-output episode illustrates
• Benefits of co-learning with an AI agent — students discovering the rules of an AI system rather than learning the discipline
• Forging ahead or proceeding with caution: Developing policy for generative artificial intelligence in legal education — policy and regulation in legal education, the same disciplinary setting

Citation

Kasa-Mälksoo, E. (2026). Challenges and opportunities of using artificial intelligence in human rights education: Reflections from higher education practice. Human Rights Education Review.