Reassembling AI-embraced Networked Learning through Actor-Network Theory

Abstract

The rapid diffusion of generative artificial intelligence (GenAI; AI) is reshaping how learning and knowledge production take place in higher education. As AI generates text, proposes logical structures, and applies academic conventions, learning can no longer be explained as interaction among human actors alone. Judgments about what constitutes a legitimate research decision increasingly emerge through entanglements among humans and nonhuman elements such as documents, institutional regulations, interfaces, deadlines, and algorithms. This shift raises foundational questions central to Networked Learning (NL) theory concerning agency, relationality, and knowledge production. While NL has conceptualized learning as a relational practice grounded in connections among learners, its core concepts have largely positioned technology as a mediating condition, leaving learning agency predominantly human-centered. AI challenges this assumption by actively intervening in meaning-making and judgment. This study aims to reconfigure key NL concepts by introducing Actor–Network Theory (ANT) as a sociomaterial analytical lens. ANT conceptualizes learning not as the outcome of human interaction alone, but as a networked effect produced through ongoing translations among heterogeneous human and nonhuman actors. Drawing on ANT concepts such as generalized symmetry, relational agency, translation, and mediation, the study reinterprets relationality, distributed knowledge creation, self-directed learning, and technological mediation within NL. Empirically, the study examines doctoral research proposal writing as a critical learning context where supervision, institutional requirements, and AI intersect. The analysis draws on phenomenological posthuman interviews with doctoral students, supervisors, and early-career researchers, supplemented by the first author’s research journal. Paired anecdotal analyses reconstruct the same research design scene through NL and ANT perspectives, enabling comparison of how different theoretical lenses foreground actors, relations, and agency. The findings show that research design decisions are not the result of individual intention or collaboration alone, but are provisionally stabilized through chains of translation involving supervisor, supervisee, documents, procedural forms, deadlines, digital systems, and AI. The study highlights how ANT makes visible the complexity of research design, repositions AI as an active translation device, and redistributes agency and power across sociomaterial assemblages. It suggests that NL research may benefit from engaging more explicitly with technology as an active participant in learning rather than as a background condition.