Redefining socio-technical systems for generative AI in education: evaluating its impact on continuance intention

Purpose Most discussions of generative artificial intelligence (AI) in education end at first-time adoption; little is known about what sustains long-term use. Drawing on socio-technical systems theory, this study reconceptualizes continuance intention by weaving together key technical affordances – anthropomorphism (human-like design cues), personalization, usability and responsiveness – with social enablers – ethical governance, AI literacy and institutional support – and tests their joint effects in an integrative model. Design/methodology/approach Survey data from 1,635 Indonesian undergraduates, lecturers, professors and research staff were analyzed using partial least squares structural equation modeling (PLS-SEM) to estimate net effects and necessary condition analysis (NCA) to uncover thresholds that must be met for continuance to occur. Findings Usefulness is the principal driver of continuance, mediated by satisfaction. Among design cues, anthropomorphism most strongly elevates usefulness, followed by personalization and usability; responsiveness adds marginal gains. Social variables (AI literacy, institutional support) weakly boost satisfaction, and personal innovativeness amplifies the usefulness–continuance link. NCA shows that ethical assurance, adaptive personalization and baseline satisfaction are indispensable: without all three, high continuance cannot occur. Originality/value The research advances policy-relevant knowledge by validating a micro-level socio-technical alignment model for generative AI, demonstrating the complementary analytical value of combining PLS-SEM with NCA. The large Indonesian sample enriches global insights into how generative AI can be institutionalized in emerging economies.