Analysis of intent to use artificial intelligence in tax assistance based on the modified UTAUT2 model

Abstract

This study aims to analyze the factors influencing taxpayers' intention to use artificial intelligence (AI) for tax assistance in Indonesia. The research design employed a quantitative approach with an explanatory survey method of 200 respondents who had used AI tools such as ChatGPT, Gemini, or similar tools for tax assistance. The UTAUT2 model was modified by adding the Trust in AI construct and removing the Use Behavior construct. Data analysis was performed using Partial Least Squares Structural Equation Modeling (PLS-SEM) using SmartPLS 4. The results showed that habit, performance expectancy, price value, and trust in AI significantly influenced behavioral intention, while effort expectancy, facilitating conditions, social influence, and hedonic motivation were insignificant. These findings confirm that intention to use AI in taxation is more influenced by practical experience and perceived value than by convenience or social influence. A limitation of this study is that it did not measure actual behavior; therefore, further research is recommended to examine this aspect. This study provides theoretical contributions to the development of AI adoption models in taxation, as well as practical implications for tax authorities and technology providers.