There’s a crack in everything – especially in ChatGPT:Analyzing the frictions within generative AI integration into a higher education context

The general accessibility of generative AI in the form of ChatGPT has opened for the educational sector to - at least partly - engage in experimentation about how to integrate this technology in everyday teaching-learning practices. Beyond overly simplistic views that either herald its advantages for personalize learning and the likes, or condemning it as unacademic, such experimentations allow for an exploration of the multi-layered nature and messiness of the situation, where educators and students alike grapple with acts of sensemaking about this pervasive technology (e.g., Vallis et al., 2023). In the wake of such experimentations, the notion arises that what the enmeshment of generative AI with (higher) education elicits are tensions (Cain, 2023), controversial sentiments (Papyshev, 2024) or even frictions (De Groot et al., 2024).<br/>Far from a merely cognitivist application of the term (Chen &amp; Schmidt, 2024), we are using the lens of frictions to analyze the messiness and underlying ambiguities that the infusion of generative AI into the educational process has brought about. For that, we are applying the feminist-inspired methodology of a frictional analysis (Puar, 2012; Staunæs &amp; Pors, 2015) to a corpus of material gathered during a small-scale educational experiment in the fall semester of 2023. A cohort of students in a pedagogical program at a Danish university engaged in an open-ended exploration of their own and their peers emerging generative AI-practices, which were documented through written assignments and during a focus-group interview. In addition, we recorded and transcribed our own discussion of these materials, thus supplementing and juxtaposing the student perspective with our respective roles as educator, educational developer and edtech researcher.<br/>By reading the material frictionally through two ostensibly contrasting theories, it becomes obvious that generative AI exists as at the same time both, an education policy object that elicits the urge to be shaped, molded and translated into the existing educational rationale (Czarniawska &amp; Joerges 1996), and a site of affective intensities (Staunæs, 2016) that activate across and liquidify the various identities and technologies involved in the endeavor of establishing learning for an AI-mediated world. <br/>