Artificial intelligence (AI) is rapidly evolving from automating routine tasks to becoming a predictive—and even prescriptive—tool in public finance. At Thursday’s New Economy Forum Workshop, two panels explored how AI and GovTech are being used across governments, and how to scale responsibly while pushing innovation forward.
“It’s not about getting one big thing right… [it’s about] getting 32 million things right,” said Edward Kieswetter, Commissioner of the South African Revenue Service. Since introducing AI tools like chatbots, biometric facial recognition for e-filing registration, and web-based assistance, South Africa has added $18 billion to its fiscal year revenue. Kieswetter pointed to three key gains: streamlining services for taxpayers, stronger compliance and fraud prevention, and most notably, increased public trust.
Across OECD countries, “there is no single or even preferred model [of adoption]”, said Delphine Moretti, Working Party Lead on Public Financial Management and Reporting for the OECD. Governments are using AI to forecast economic trends and help inform spending decisions. France and Indonesia, for instance, use AI to monitor fiscal risk at the subnational level through accounting data. Still, oversight bodies, public financial management frameworks, and communities of practice are critical to help manage risk and ensure that innovation leads to real gains.
In Brazil, AI is also being leveraged for fiscal education. Tania Gomes, Coordinator for Data, Products and Digital Transformation, Treasury of Brazil, showcased “Talk to SICONFI”, a generative AI agent that answers queries on public fiscal data across federal, state, and local levels. Promoting training and digital literacy for AI is just as essential, she added.
AI tools can be scaled broadly at extremely low costs, but doing so requires strong risk management frameworks and agile governance, says David Hadwick, a researcher at the Centre of Excellence ‘Digitax’. Spanish Tax Agency’s Chief Information Officer, José Borja Tomé, illustrated this with the agency’s “test-and-pause” approach, underscoring that “assigning responsibility is key”.
Panelists agreed that policies guiding AI use in public finance should prioritize transparency, fairness, efficiency, and use trusted, high-quality data. Increasingly so, “the metrics of AI ethics correspond to the metrics of performance for these administrations,” Hadwick added.
Culled from IMF.org