Inteligência artificial generativa na auditoria governamental: estudo de caso do ChatTCU no Tribunal de Contas da União

Abstract

Digital transformation in the public sector has driven the adoption of emerging technologies to enable greater efficiency and transparency in public management. In this context, Generative Artificial Intelligence emerges as a technology with the potential to optimize complex processes, including government auditing. This research aimed to investigate the factors influencing the acceptance of the ChatTCU solution by auditors of the Brazilian Federal Court of Accounts (TCU), based on the application of the Unified Theory of Technology Acceptance and Use (UTAUT), evaluating the following predictors: performance expectancy, effort expectancy, social influence, enabling conditions, trust and attractiveness of alternatives. ChatTCU is a virtual assistant based on generative AI to assist auditors in audit procedures and process preparation. However, the effective incorporation of this technology into government auditing activities depends on its acceptance by auditors, which motivated this research. The methodology adopted used a quantitative and descriptive approach, applying a structured survey to TCU auditors located at the agency's headquarters and regional units. The collected data were analyzed using exploratory factor analysis, multiple linear regression with bootstrapping, and the general linear model (GLM). The results showed that performance expectancy and trust are the factors that most strongly influence auditors' intention to use ChatTCU, while social factors and attractiveness of alternatives also have a positive influence, albeit more moderately. The research findings allow the TCU and other courts of accounts that also use the ChatTCU solution to better direct their efforts and strategies to expand the acceptance of AI technologies within the organization.