How Generative AI Competence Shapes Student Engagement: Configurational Pathways Through AI Utilisation,Perceived Autonomy,and Formal AI Learning

Abstract

This study moves beyond linear models to examine how GenAI competence, utilisation, perceived autonomy, and formal AI learning combine to shape student engagement. The research process involved collecting data from 262 university students via convenience and snowball sampling, applying a configurational approach grounded in Complexity Theory, and performing fuzzy-set Qualitative Comparative Analysis (fsQCA). Analytic steps included calibration to set-membership values, necessary condition analysis, truth table construction, derivation of configuration solutions, and Tobit regression. The analysis identifies three pathways to high engagement: (a) Compensatory Pathway (strong AI utilization compensates for lower formal learning when autonomy is present); (b) a Self-Directed Pathway (high autonomy and informal utilization suffice regardless of formal training); and (c) the Comprehensive Pathway (all four conditions work in concert). All three solutions significantly predicted engagement (p < .001), with coefficients of 1.011, 1.513, and 1.279. A core finding is that GenAI utilisation is indispensable across all pathways. No single condition was necessary, justifying engagement as an emergent, configuration-dependent outcome. Validated through robustness checks, these configurational insights reveal multiple equifinal routes to high engagement, offering a nuanced framework for designing tailored, equitable interventions in AI-enhanced learning settings.

Clinical Trial Number: Not applicable