Generative AI Adoption by Academic Researchers: An Integrated UTAUT–Social Exchange Model Using SEM-ANN-IPMA Analysis

Abstract

Generative artificial intelligence (GAI) is changing how scholarly knowledge is produced and managed, yet the researchers who form the core of the knowledge workforce remain hesitant adopters. Higher education institutions must decide how to govern, resource, and support GAI use across the research process. Existing acceptance research concentrates on students and classroom use, treats the technology as a passive tool, and gives little attention to the costs researchers weigh against its benefits. This study models adoption as an exchange between the researcher and the tool, integrating the Unified Theory of Acceptance and Use of Technology (UTAUT) with Social Exchange Theory (SET). Eleven hypotheses, grouped into benefits, costs, and contextual factors, were tested with artificial intelligence literacy as a control; the constructs were specified from a published systematic review and expert prioritization study. Survey data from 238 academic researchers at fourteen higher education institutions were analyzed with partial least squares structural equation modeling, extended with importance–performance map analysis and an artificial neural network to rank the drivers for decision-makers. The model explains 57.3% of the variance in behavioral intention. Knowledge acquisition is the strongest driver, followed by researcher–GAI reciprocity and task efficiency; perceived workload lowers intention, cognitive risk shows a weaker negative effect, and institutional support shows none. Adoption depends on the quality of the exchange, meaning what the tool returns in knowledge and what it demands in effort. This gives institutions an evidence-based basis for managing GAI in the research workforce.