The Unintended Consequences of Artificial Intelligence and Education

The Unintended Consequences of Artificial Intelligence and Education is the research report authored by Dr Wayne Holmes and published by Education International on 18 October 2023. The report provides a systematic analysis of the current state of artificial intelligence in education — its potential benefits, its documented risks, and the role of teachers and teacher trade unionists in ensuring that AI in education is aligned with the principles of social justice and human rights. It is the flagship analytical publication in Education International's sustained campaign to position global education unions as active shapers of AI's trajectory in education, not passive recipients of technology decisions made elsewhere.

Education International: publisher and advocate

Education International was founded on 26 January 1993 from the merger of two predecessor global bodies — the World Confederation of Organisations of the Teaching Profession (WCOTP) and the International Federation of Free Teachers' Unions (IFFTU) — and is now one of the world's largest sectoral global union federations, with 383 member organisations across 178 countries and territories representing over 32 million education workers from pre-school through university. Headquartered in Brussels and holding UNESCO associate status and UN ECOSOC consultative status, it functions as the authoritative global voice of organised educators on policy questions touching teaching, learning, and the governance of education systems. General Secretary David Edwards has led the organisation since 2018; Mugwena Maluleke was elected President in 2024.

EI's institutional position on AI is that it carries both transformative potential — for teaching, learning, research, and union organising — and serious systemic risks: left unchecked, AI "can undermine education as a human right, erode educators' working conditions and professional autonomy, and perpetuate harmful social and environmental consequences." The Holmes report is the analytical engine behind that institutional position. EI's prior major publication in the same research tradition was Teaching with Tech: The Role of Education Unions in Shaping the Future (2020), authored by Dr Christina J. Colclough — released during the COVID-19 school-closure period when EdTech adoption surged globally and establishing an EI research tradition that treats AI not merely as a pedagogical question but as a governance and worker-rights question requiring union engagement.

Author: Wayne Holmes

Wayne Holmes is an Associate Professor at UCL's IOE, Faculty of Education and Society, where his work examines the teaching and application of AI in educational contexts through lenses of ethics, human rights, and social justice. He leads the Council of Europe's expert group on AI and Education, which is developing human rights protections for students and teachers using AI systems, and serves as a consultant for UNESCO's Technology and AI in Education unit and a senior researcher at IRCAI (International Research Centre on Artificial Intelligence under UNESCO). His prior publications include AI and Education: Guidance for Policy-makers (UNESCO), Guidance for Generative AI in Education and Research (UNESCO), and co-edited volume The Ethics of AI in Education: Practices, Challenges and Debates (Routledge, 2022). His appointment to write this report bridges the intergovernmental-standards world (Council of Europe, UNESCO) and the labour-movement world (Education International) — the report is analytically grounded in Holmes's academic and advisory work but oriented toward trade-union practitioners, not policymakers.

Structure and scope

The report organises its analysis around three deployments of AI in education: student-focused tools (adaptive tutoring, chatbots, virtual assistants, automated essay scoring); teacher-focused tools (AI for lesson planning, professional development, administrative workload); and institution-focused tools (automated proctoring, admissions screening, learning analytics dashboards). The three-part structure maps AI's incidence on all three primary actors in an education system rather than treating AI in education as a single phenomenon, and it reveals that the risks are structurally asymmetric: student-facing AI has received the most commercial attention and the most inflated claims, teacher-facing AI remains largely underdeveloped relative to those claims, and institution-facing AI operates largely out of view of the people it most directly affects.

Central findings and critique

The report's critical core is that claims made for AI in education — especially for personalisation and adaptive learning — are substantially unsupported by independent evidence. Studies on student-facing AI tools are few, methodologically narrow, and disproportionately funded by commercial stakeholders with a stake in the conclusions. Where AI for students is commercially deployed at scale, student data tends to be commercially appropriated, and personalisation in practice tends to mean automating the route to a pre-fixed educational destination rather than expanding what education can achieve. Holmes makes this structural critique concrete through a transport metaphor: AI, as currently deployed in education, makes the journey more efficient but does not change the destination — "if the destination doesn't change, I may as well take the bus" — and in doing so it perpetuates existing systems rather than transforming them.

Teacher-facing AI is found to be largely underdeveloped relative to commercial claims, while institution-facing AI — automated proctoring, recruitment screening — operates in ways largely invisible to students and teachers, concentrating decision-making power and accountability risk in institutional and commercial actors without corresponding transparency. The report identifies an expertise gap cutting across all three categories: computer scientists understand AI deeply but education superficially; educators understand education but lack the technical fluency to interrogate AI systems — a mismatch that systematically advantages commercial AI vendors in procurement decisions and policy debates. The three core systemic risks that flow from this configuration are AI reducing educational quality, deteriorating working conditions for educators, and producing substandard learning experiences for students.

Recommendations and the union role

The report's recommendations orient around two complementary principles: "ethics by design" — the principle that ethical considerations must be embedded in AI systems from the development stage rather than bolted on retrospectively — and AI literacy as a prerequisite for effective teacher and union agency. Teachers and teacher trade unionists are called on to critically evaluate AI tools before adoption, to demand transparency from developers about how systems work and what data they collect, and to develop AI literacy that encompasses the social, pedagogical, and political dimensions of AI deployment alongside the technical. The union dimension is explicit: EI's framing positions collective bargaining, not just individual professional judgment, as the appropriate governance mechanism for AI's entry into education systems.

This positions the Holmes report alongside the UNI Global Union Top 10 Principles for Ethical Artificial Intelligence (2017) in the global labour-movement literature on AI governance: both publications treat AI governance as a workers' rights question requiring organised-labour capacity-building, not simply an ethics or regulatory question. The EI report is written nearly seven years later, in the GenAI era, and the shift is visible — from abstract principles to a demand for transparency and evidence in specific deployed tools, and from sectoral generality to education-specific analysis of the three-layer architecture that AI in education actually takes.

EI subsequently convened its inaugural Global Conference on Artificial Intelligence in Brussels on 4–5 December 2025, under the theme "Shaping our Future: Education Unions Leading for a Human-Centred AI," gathering over 200 union leaders, educators, and experts from all regions. The conference built on EI World Congress resolutions calling for a human-centred, ethical, and rights-based approach to AI in education, and aimed to equip unions with concrete strategies for AI governance and organising at national and regional levels — operationalising the 2023 report's call for union capacity-building into a global mobilisation frame.

Position in the corpus

The Unintended Consequences of Artificial Intelligence and Education is the corpus's first education-sector publication anchor. It closes the sectoral-advocacy coverage gap for education identified by the Synthesizer: the corpus holds substantial coverage of the AI-rights movement across legal advocacy (Foxglove, Privacy International), digital rights (Access Now, EFF), worker and platform organizing (the WGA-SAG-AFTRA AI provisions, the UNI Global Union principles), disability (European Disability Forum), humanitarian-disarmament (Stop Killer Robots), and geographic civil society anchors across multiple world regions, but lacked any publication by or for the education sector despite education being one of the primary societal domains where AI deployment is contested and actively organised. Education International's position as the world's largest education union federation and the Holmes report's explicit trade-union framing make this the natural first education-sector anchor — it is a union-commissioned analysis whose primary audience is teacher organisers and their members, and whose recommendations are framed in the language of worker rights, collective bargaining, and democratic accountability rather than institutional policy, placing it squarely in the corpus's grassroots and small-d democratic layer.