Conference Program

Education governance is increasingly reorganised through artificial intelligence, data infrastructures and automated decision-making. Instead of treating AI as an instrument that enhances human policy rationality, this paper adopts a more-than-human approach to analyse how platforms, algorithms and data reconfigure agency, cognition and accountability across education systems. Building on Science and Technology Studies and new materialisms (Barad, 2007; Lemke, 2021; Massoumi, 2002; Whatmore, 2013) and education policy research on datafication and computational governance (Williamson, 2017; Gulson, Sellar & Webb, 2022), we operationalise three analytical tools (post-anthropocentrism, symbiosis and affectivity) to examine how AI participates in governing beyond the human.

Contemporary policy debates remain largely framed through human-centric categories such as fairness, efficiency and transparency (Selwyn, 2022), which presume that technology either supports or distorts human judgement. A more-than-human perspective instead treats governance as emerging from entanglements of humans, datasets, infrastructures, legal dispositifs and computational forms of cognition (Parisi, 2019; Hayles, 2014). AI does not simply execute policy decisions, instead it generates classifications and predictive logics that shape institutional futures and subjectivities. In this sense, governance is enacted through sociomaterial assemblages rather than located in human actors (Landri, 2018).

We deploy three empirical vignettes from England and Italy illustrate this shift, examining first the 2020 A-level grading algorithm in England, which was designed to standardise results after exam cancellations but disproportionately downgraded students from historically underperforming schools, thus reproducing structural inequalities embedded in training data (Williamson & Piattoeva, 2019).

In a second vignette we analyse the Italian algorithm allocating temporary teachers through automated ranking. This algorithm was introduced as a solution to bureaucratic inefficiency, but the system produced mismatches, labour insecurity and legal disputes. The third vignette instead contrasts these extractive configurations with IAQOS, a community-trained neighbourhood AI developed in Rome (Iaconesi & Persico, 2021). Through participatory design and open-source infrastructures, residents and students collectively shaped the system’s knowledge base.

Across the three cases, AI governance appears as a cognitive-affective formation rather than a neutral technical layer. Algorithms mobilise affects, including anger, anxiety, trust and attachment that reorganise institutional legitimacy and democratic participation. A more-than-human approach does not reproduce techno-solutionism instead it provides analytical tools to map heterogeneous agencies and to distinguish extractive assemblages from participatory and convivial ones. In doing so, it contributes to Critical EdTech Studies by reframing AI governance as a democratic and relational problem, aligned with ongoing debates on datafication, privatisation and the futures of education policy (Selwyn, 2022; Williamson et al., 2024).