| Sumario: | Background: As generative artificial intelligence (GenAI) becomes more deeply integrated into AI‐mediated informal digital learning of English (AI‐IDLE), understanding how learners organise their acceptance of these tools is increasingly important. Existing research has largely relied on variable‐centred approaches, offering limited insight into how acceptance beliefs are configured across learner groups. Objectives: This study examines how learners' acceptance of GenAI beyond the classroom is structurally organised and how these configurations vary across educational levels and disciplinary backgrounds. Methods: Grounded in the Integrated Model of Technology Acceptance (IMTA), the study employed Epistemic Network Analysis (ENA) to model four acceptance networks: overall IMTA, perceived enjoyment (PE), perceived usefulness (PU) and negative use experience. Semi‐structured online interviews were conducted with 24 Chinese university students (BA, MA, PhD; humanities and social sciences, STEM) and theory‐driven coding was used to construct and compare network structures. Results and Conclusions: Findings revealed a developmental reconfiguration of acceptance. BA learners' IMTA networks were experience‐oriented (PE, PEU), whereas postgraduate learners showed more utility‐driven configurations integrating PU and behavioural intention. PE networks showed disciplinary differences and some developmental variation, shifting from accompaniment‐centred structures towards confidence‐oriented patterns. PU displayed the clearest educational differentiation, progressing from affordance‐based evaluations to goal‐aligned and critically engaged use. In contrast, negative‐use networks showed structural stability across educational levels but differed by discipline. Overall, GenAI acceptance in AI‐IDLE emerges as a developmentally structured and motivationally layered process rather than a static set of beliefs. Practitioner Notes: What is already known about this topic ○GenAI is increasingly used in AI‐IDLE.○PE and PU predict learners' intention to use GenAI beyond the classroom.○Most studies use survey‐based models (e.g., TAM/IMTA) focusing on linear relationships.○Disciplinary and educational differences in GenAI acceptance remain underexplored.What this paper adds ○Acceptance in AI‐IDLE is structurally organised rather than a set of isolated beliefs.○Educational level shapes how PE and PU are configured within acceptance networks.○STEM and HSS learners differ in affective and negative perception structures.○Negative perceptions remain stable across levels but vary by discipline.Implications for practice and/or policy ○Support for undergraduates' AI‐IDLE should emphasise affective engagement and ease of use.○Postgraduate training should promote critical and goal‐aligned GenAI use.○Disciplinary contexts should inform how GenAI integration is designed.○Risk awareness and critical evaluation should be embedded in AI‐assisted education.
|
|---|