In an innovative leap forward, Armenia is piloting the use of artificial intelligence (AI) to strengthen tax compliance, risk management, and fraud detection, supported by the World Bank. On April 17, 2025, Armenia’s State Revenue Committee showcased its AI-powered tax administration tool and signed a Memorandum of Understanding with the American University of Armenia for their capacity support on AI.
As one of the first movers among developing nations, Armenia’s experience offers key lessons for governments exploring AI in public service. Here are the top takeaways from the journey so far:
1. Leadership is the Game Changer
Success begins at the top. In Armenia, reform was led by the Chair of the Revenue Committee, who focused on solving real business problems—not chasing technology trends. This high-level commitment and problem-first mindset were critical in driving impactful change.
2. Bridge the Capacity Gap Through Partnerships
Limited technical capacity is not a binding constraint. Armenia addressed this challenge by partnering with academia and the private sector. With World Bank support in financing, convening, and knowledge sharing, the State Revenue Committee tapped into the American University’s data science expertise—giving researchers and students hands-on exposure while expanding government capability.
3. Hiring AI Talent? Think Smart Collaboration
AI expertise—spanning data scientists, architects, and developers—is scarce and expensive. Armenia sidestepped hiring bottlenecks by working with the World Bank to design competitive procurement processes. Partnering with the American University created a win-win setup: real-world projects for students, and much-needed skills for the government.
4. Focus on High-Impact Use Cases
Forget flashy tech—focus on solving problems. Armenia aimed at improving tax compliance, a top priority for many governments. The AI solution uses natural language processing to read invoices, network analysis to detect fraud rings, and predictive modeling to flag anomalies like duplicate filings and identity mismatches. Beyond internal use, administration is considering citizen-facing applications: AI chatbots can answer tax queries and AI tools speed up tax refund processing. Both offer direct public value.
5. Manage the Risks Early
AI is not without risks — bias, data privacy, security, and job displacement. Armenia addresses these risks by focusing on four pillars: explainability (using clear, non-technical logic), transparency (opensource code), strong cybersecurity (aligned with ISO standards), and stakeholder engagement (involving tax firms, media, and civil society in oversight). Algorithm Impact Assessments in future could be carried out to uncover bias.
6. Cut the Hype
AI is everywhere, but it is not always needed. Armenia avoided the trap of tech-for-tech’s-sake by sticking to measurable, problem-driven goals. Real success lies not in headlines, but in results. Armenian government expects the use of AI to be very impactful - better compliance rates (15-20 percent at least), reduced administrative overhead, and more consistent revenue flow.
To conclude, Armenia’s experience shows that even small nations can lead big changes. With strong leadership, smart partnerships, and a clear focus on outcomes, AI in tax administration isn’t a distant vision—it’s happening now.
