<p>Governments are moving to embed artificial intelligence (AI) in the core machinery of fiscal governance, which may affect distributive decisions regarding who receives funding (and who doesn’t), and institutional viability of algorithm-enabling budgeting reforms depends largely on whether they are viewed as preferable by the bureaucrats who are charged with implementing them. While the public administration literature has given considerable attention to the efficiency gains associated with AI, far less is known about how civil servants themselves respond to the prospect of adoption of AI for budgetary decision-making, which extends beyond questions of administrative productivity and enters the domain of distributive governance. This study tests bureaucratic support for AI-assisted budgeting relative to alternative reform pathways: workforce expansion and participatory budgeting. We implemented a three-arm survey experiment among a large sample of public sector personnel (<i>N</i> = 3,820), randomly assigning respondents to vignette scenarios describing each reform. Manipulation-restricted robustness analyses, weighted to account for differential treatment recall, show that AI-assisted budgeting reforms present significantly higher bureaucratic support than workforce expansion, while presenting no statistically significant differences relative to participatory budgeting treatments. Intent-to-treat estimates with covariate adjustment and organisational fixed effects do not show statistically significant differences across outcome dimensions.</p>