The Extended UTAUT Model in Predicting ChatGPT Adoption: Analyzing the Influence of Performance Expectancy, Effort Expectancy, Self-Efficacy, Social Influence, and Facilitating Conditions on Actual Use by Biology Education Students

Abstract

Background: Generative Artificial Intelligence (GAI), particularly ChatGPT, has emerged as a revolutionary learning tool that can improve academic efficiency in Biology education. However, the factors driving student acceptance in this discipline need to be explored further.Aims: This study integrates the Unified Theory of Acceptance and Usage of Technology (UTAUT) model with the Self-Efficacy (SE) variable to investigate the intention and behavior of using ChatGPT.Method: Using a quantitative approach, data were collected from 235 valid participants enrolled in Biology Education programs across two public universities in Indonesia and analyzed through partial least squares structural equation modeling (PLS-SEM) using SmartPLS software.Results: Performance expectancy, social influence, and self-efficacy significantly influence behavioral intention, with self-efficacy as the strongest predictor (β = 0.392, p = 0.000). Conversely, effort expectancy and facilitating conditions have no significant influence on behavioral intention (R^2=0.708). Critical findings reveal a severe gap between intention and action, where behavioral intention fails to significantly influence actual use (β = 0.075, p = 0.471), supported by a very low R^2 value for actual use (0.006).Conclusion: This significant gap indicates that contextual barriers such as academic ethics policies prevent students' intentions from translating into actual behavior. Implications are provided for educators to improve AI literacy and align policies to optimize the integration of GAI in Biology education.