Interrogating the Hyperbole Machine: A Text-Mining Approach to IOs, AI and Education Policy Frameworks

• In Event: Global Governance, Transformative Relations, and Human Rights in Education

Proposal

This paper interrogates how international organizations frame AI literacy, understood as an emerging socio-techinal assemblage with a distinct political economy (Birch 2025) giving rise to new political orders (Amoore, 2023). It does this by analysing three key policy documents: the OECD’s Empowering Learners for the Age of AI (2025), UNESCO’s AI Competency Frameworks for Students, and UNESCO’s AI Competency Frameworks for Teachers (2024). To this aim, we employ data-mining techniques and network text analysis to pursue two interrelated analytical trajectories. First, we identify the main actors, map their mutual interactions, and examine the forms of agency ascribed to them by the international organizations under consideration. Second, we trace the emergence of two key terms arising from these policies, ‘AI’ and ‘human’, and explore how these concepts become the ideological poles of efforts to radically reconfigure social relations in the educational landscape.
Our analysis reveals significant similarities and differences both between the analysed policy texts and, diachronically, between them and previous policies on edtech. The first line of analysis highlights an almost exclusive focus on AI and students as the main actors in the AI society, while the role of teachers and educational institutions is significantly sidelined. Within this landscape, only the UNESCO teachers’ document stands out as attempting to define the role of teachers in the new AI society. Subsequently, the second line of analysis indicates a significant ideological shift that encompasses all the studied documents. In these policies, “AI” is portrayed as ubiquitous and all-mediating, compounding with every aspect of education–society relations, while the “human” is framed dialectically, again around a series of compound relations.
Taken together, these two transformations point to a broader dramatic socio-technical shift: rather than adding technology into society, these documents depict society as being added to AI; a necessary variable that continuously reshapes both its technical functioning and social purpose. Within this framework, students are positioned as directly and unrestrictedly engaging with AI systems, becoming “cybernetic” producers of knowledge, while teachers are largely marginalized, particularly in the OECD’s framework. In short, education becomes what Marres (2025) elsewhere has described as a testing ground for AI systems more generally. We show that traditional educational actors (e.g., teachers and institutions) are tasked with providing state legitimation and basic monitoring, but not with interfering. We argue this represents a profound departure from traditional edtech policy, where incremental adaptation following testing and scaling up has been typical. Here, AI is leveraged as a totalizing force, reshaping educational priorities around testing and productivity whilst erasing/obscuring questions of agency, critique, teacher mediation and institutional context.
References
Amoore, L. (2023). Machine learning political orders, in Review of International Studies, 49 (1) 20-36.
Birch, K. (2025) Do artifacts have political economy? Science, Technology and Human Values, https://doi.org/10.1177/01622439251352167
Marres, N. (2025). How tech trials put society to the test, Theory, Culture and Society, in https://journals.sagepub.com/doi/10.1177/02632764251342911.1177/02632764251342911
OECD (2025) Empowering Learners for the Age of AI, Paris: OECD.
UNESCO (2024) AI Competency Frameworks for Student, Paris: UNESCO.
UNESCO (2024) AI Competency Frameworks for Teachers, Paris: UNESCO.

Author

• Susan Lee Robertson, The University of Manchester