Opinion: AI at the tax office: A powerful idea but not a quick fix
AI in the Belgian tax administration needs responsible foundations, accessible services and human-centred design to create real value.
The Belgian government recently announced that the tax administration will begin using AI to detect tax fraud more quickly. As an UX Lead I can certainly appreciate this. But what does this mean for the user, for society and for the future of digital services?
Because, frankly, many of the government’s core digital services still feel stuck in the early 2000s.
Directly after I read the article in “De Morgen - The tax authority will use AI to detect tax fraud more quickly.” I found myself questioning how the Belgian government will handle this knowing that the average citizen still struggles to navigate existing platforms like MyMinfin, eBox and health insurance portals. I understand that we can’t wait to implement AI until everything is user-friendly. But the art is not to do one before the other, but to build both fundamental layers simultaneously.
AI must be responsible and usable
Introducing AI in the public sector is not just about technology. It demands clear fundamentals, built on ethics, accessibility, transparency and sustainability. That’s a struggle a lot of average citizens deal with today.
In my role at Sopra Steria, where I’m a UX Lead and People Manager Design, I see everyone chasing faster delivery and leaner teams. It’s a race driven by competitive pressure and tight budgets. But it’s important that speed does not come at the expense of transparency and understandability.
The real value for users
This is why UX fundamentals and best practices matter more than ever. We can’t just assume that AI alone will be the solution. Without proper research, testing and proven best practices, AI will risk becoming just another complex layer that users don’t understand or trust.
The same principles we’ve always applied in UX s.a. listening to real users, mapping their journeys, ensuring accessibility, making flows transparent and predictable are still the foundation for meaningful AI adoption.
Because if those basics aren’t there, we’re simply automating frustration. A badly designed flow won’t suddenly become clear just because an algorithm powers it. AI can only create real value for end users when it is built on top of solid, human-centered design practices.
Our approach: Building AI, with the right foundations
At Sopra Steria, we don’t see AI as just another tool to roll out. For us this means creating responsible AI frameworks before deployment. We build AI-powered features with proven best practices so it’s usable, accessible and trustable for everyone.
We apply this across the company, providing role-specific training that goes beyond technology and tools. We measure success by the meaningful outcomes it creates, not just by efficiency.
At Sopra Steria, that’s where we’re placing our bet. Not only accelerating what already exists but on fundamentally redesigning the systems themselves.
Originally published on LinkedIn on 16 October 2025.