The rapid integration of Generative Artificial Intelligence (GenAI) into higher education has sparked a search for order amid significant upheaval. Current academic discourse often relies on "hasty," unreflective role metaphors – modeling GenAI as a "learning partner" or "corrector" – to make the fundamental uncertainty (contingency) of the technology manageable within familiar pedagogical frameworks. However, this paper argues that such translations create a theoretical gap, attempting to describe radical socio-technical shifts using the vocabulary of the old. To bridge this gap, we apply perspectives from cultural and practice theory alongside Actor-Network Theory (ANT). We define GenAI as a "contingency-generating" technology that produces "agentive contingency," continuously generating unexpected results that disrupt established "cultures of competence". By adopting a flat ontology and the guiding principle to "follow the actors," we move beyond the subject-object separation to explore how agency is distributed within hybrid human-machine associations. A case study of linguistic interaction illustrates how GenAI functions as a "black box" that shifts the hierarchy of competence from intellectual grasp to the ability to address technology for situational respecification. We conclude that future- proof education must: Shift the pedagogical focus from automatable results to the complex process of posing questions; Engage in "un-boxing" by making the socio-technical network's influence visible; Preserve the university as a contingency-sensitive space where the fluidity of existing orders is used as a productive moment for transformation.