Applications of Generative Artificial Intelligence in Tax Practices: Evidence from Tax Professionals
基于18位税务专业人士的访谈，研究了生成式AI在税务实践中的实际用途、价值与局限，发现它主要作为辅助工具支持法规解读和探索性分析，而非替代专业判断。
SYNOPSIS This study examines how generative artificial intelligence (Gen AI) is currently used in professional tax practice and how tax professionals assess its value and limitations. Drawing on 18 semistructured interviews with tax and tax-technology professionals, we document that Gen AI is already supporting core tax activities such as research, issue identification, workpaper preparation, and compliance diagnostics. We find that Gen AI is primarily used as a reference and copilot tool to support statutory interpretation and exploratory analysis, rather than to automate tax judgments or replace professional expertise. Although adoption is driven by efficiency and competitive pressures, tax-specific concerns, such as accuracy, documentation, regulatory scrutiny, and liability, constrain its use. Consequently, participants emphasize human oversight and restrict autonomy. Participants anticipate Gen AI to reshape how tax work is performed rather than to displace tax professionals. These findings provide timely insights into the evolving role of Gen AI within the tax function. Data Availability: Interview transcripts are not publicly available due to participant confidentiality agreements, but anonymized excerpts supporting the findings are included in the manuscript. JEL Classifications: M4; H2.