The Innovantage podcast is here not just to inform you about some technological advancements but to provide you with a comprehensive vision of how AI-powered digital government can be integrated into our daily lives and the world around us.
In the 10th episode, Max Golikov, the podcast host and Sigli’s CBDO, invited for a discussion Hans Arents, a senior advisor for AI-powered digital government at Digitaal Vlaanderen.
At the beginning of his professional life, Hans gained expertise in the private sector and further worked as an AI researcher at the university. Now, for almost 25 years, he has been working for the Flemish government.
Max and Hans talked about the potential of AI to revolutionize multiple processes across the public sector as well as about the challenges that currently exist.
It’s interesting to mention that one of the previous Innovantage episodes provides insights into how the Estonian government is undergoing its transformation phase. Now, our readers (or viewers, if you prefer a video format) have the possibility to compare two approaches.
The Flemish Government’s AI Approach
When Hans joined the Flemish government, there was a global plan to build an e-government. Today this concept is already in the past. Now, we can observe a shift toward the idea of digital government. This is a broader concept, which presupposes that it is not enough just to transform some separate processes. It is much more important to become digital entirely.
Some promises made in 2000 still haven’t been fully fulfilled. It happened because of several factors. First of all, the government is rather slow in its decisions and it has to be very careful about all its initiatives. Secondly, there is an obvious mismatch between the speed at which the government can move and the speed at which technology moves. This makes it difficult to stay up to date with the latest technological developments, especially in AI.
But despite this, the Flemish government fully realizes that the digitalization of its services and internal operations is the key to achieving better operational efficiency and better service delivery to its citizens and enterprises. And today AI is viewed as one of the main technologies to support this transformation.
The history of the AI introduction in the Flemish government is just beginning. However, the government has already conducted some experiments with artificial intelligence, machine learning (ML), and deep learning (DL). These experiments demonstrated a real need to integrate AI into the government’s operations.
Flemish government’s digitalization initiatives
In 2006, the MAGDA platform was introduced and for last years it was the heart of the Flemish government’s digital transformation. It is a data exchange platform that serves as a backend for delivering digital services to citizens and enterprises.
It is not powered by AI. It has a service-oriented architecture and it facilitates the process of working with data from authentic data sources (registries).
All the information that citizens or enterprises share with the government once is kept on databases and they do not need to give this information again and again. The government can just reuse the same data for multiple purposes.
Now, thanks to ML, DL, and generative AI, the government can significantly improve its internal operations. To achieve its goals, an AI expertise center was established. It is responsible for drafting an AI strategy and now it has four different goals:
- To prepare the Flemish administration (and especially civil servants) for the use of AI;
- To stimulate innovation using AI;
- To ensure a trustworthy use of AI;
- To build a number of reusable AI components so that it will become easier for the Flemish government and the local governments to deliver AI-based data-driven solutions.
Unique way of the Flemish government in AI implementation: Digital-first but not digital-only
For the development of its AI strategy, the Flemish government also studies the examples of other countries. For example, Hans shared that they like the idea of the Bürokratt chatbot in Estonia. This solution is designed to help citizens and enterprises get access to all public services through one communication channel.
The Flemish government now has plans to introduce a Bürokratt-inspired AI-powered digital assistant. This solution is expected to be implemented within the next five years.
Among other countries whose experience is under consideration, Hans named the UK and Netherlands.
However, Belgium in general and the Flemish region in particular have some unique challenges in implementing digital services.
The country has a federal government, three regional governments, and multiple local governments. And the services that citizens may need to use are managed by different authorities at different levels. That’s why a lot of integration efforts are required. Here, Hans talked not only about technical integration but also legal and organizational integration. Otherwise, it will be impossible to ensure seamless service delivery.
Moreover, Hans admitted that digital literacy in Belgium is around 70% which is significantly lower than in Estonia, where this figure is around 87%.
The digital gap existing within Belgian society is a serious problem of digital inclusion. Given this, Belgium needs to implement a digital-first approach, but not digital-only.
The provision of personalized digital services requires the use of even more personal data. To do it, the government always needs to ask citizens for permission to do it. And to give a grounded answer, people should have a great awareness of what data the government has about them and what the government can and cannot do with that data.
Given all this, the Flemish government has created a so-called data utility company Athumi. This company provides digital data wallets or data vaults, in which citizens can store data. And the government or private enterprises can access them and use that data but only if they have received explicit consent from citizens.
According to Hans, this model will be used for providing personalized services.
Six principles for the use of AI in government
While talking about what is required from the government to be ready to use AI, Hans named the main requirements for it:
- To be democratic;
- To be trustworthy;
- To be human-centered;
- To be sustainable;
- To use data properly;
- To respect all the legislation, which is defined at the European level.
And among these principles, Hans named trustworthiness as the key one.
AI regulation: Net positive or net negative?
The European Commission has introduced the AI Act. According to Hans, it happened soon enough to prevent any serious bad consequences of using artificial intelligence without proper control. This act includes more than 200 pages of very densely written regulation. And the Flemish government is yet to understand how this regulation will affect the offered services. The European Commission has always said that according to their estimates, only about 10-15% of all AI systems will fall in the high-risk category.
However, in the public sector, the majority of systems will fall in this category. It means that a lot of effort will go into meeting the newly introduced requirements.
Though strict regulation may be viewed as a rather controversial thing when it comes to innovations, Hans believes that in the case of AI, it is more positive than negative. At least now those who want to innovate will know what principles to respect and what restrictions exist.
For example, in the US, there is no AI regulation but there are hundreds of different laws. Such a situation creates an absolutely confusing environment.
Hans also compared regulation with a mother-in-law of innovations, which means that it’s better to have a good relationship with it if you want to innovate.
Ethics and AI
Hans also mentioned that in their AI strategy, they also defined a number of ethical principles. It is vital to respect them and follow them. However, it will be the hardest thing to check before you start an AI project whether you meet all the requirements indicated in these ethical principles.
The Flemish government is considering the possibility of creating an ethical board consisting of external experts. They will help to check whether the developed AI systems are compliant with the fundamental ethical principles.
This ethical aspect greatly differs for private and public sectors. The ethics that a company has chosen may be based just on the idea of providing value to its shareholders. And that will be enough.
For governments, it doesn’t work this way. Governments also have to respect the fundamental human rights and the dignity of people. Everybody should be treated the same way. For a private company, it is not essential to do all this.
What is perfectly acceptable for a company may not be acceptable for a democratically elected government.
How the Flemish government plans to use AI
The long-term goal is to have a government that will be smarter, more agile, and more proactive. The word smarter in this context means that the government should make better data-driven decisions. Today the government has a lot of data that was collected but that is not used for making decisions. As for agility, the government should be able to react far faster to new developments.
A nice example of proactive services that the Flemish government offers already now is the child support program. When a child is born, citizens do not need to ask for child support. It will be provided automatically. When a child goes to primary school, this support becomes higher. But again, citizens do not need to provide any extra information. Everything is done automatically.
Now, to increase the personalization of service delivery, the government plans to apply AI.
When it comes to service delivery, Hans noted: “We want AI to stand not for artificial intelligence. We want AI to stand for adaptive interactivity.”
It means that when citizens need to get access to some services, they don’t have to visit multiple portals. They don’t even have one single portal. They have personal portals. It means that what they see on the portal is different from what their neighbors may see on theirs.
Such a portal should be so hyper-personalized that people will see only those things that are relevant to them and to their level of digital expertise.
This project will be possible only with AI.
Data-driven policymaking and AI-powered simulations
Hans mentioned two key aspects of how AI can be applied to data-driven policy.
Governments have a lot of useful data. And politicians, instead of deciding based on their intuition, should base their decisions more on data-driven insights.
The second use case where AI plays a huge role is the simulation or creation of so-called digital twins of some regions or cities in different situations. For example, thanks to them it is possible to see what is going to happen if certain streets are closed for cars, what is going to happen to the air pollution in other streets, what impact it may have on public transport, etc.
When governments want to prepare for some climate-related disasters, they can also simulate them.
The existing AI initiatives
The Flemish government has launched the “amai!” initiative which is aimed at engaging citizens in the process of building new AI apps. People can offer their own ideas, evaluate the existing suggestions, and share their feedback.
The more people know about AI, the fewer concerns and fears they have related to this technology.
The Flemish government is trying to apply AI in four domains:
- Day-to-day backend work of civil servants;
- Communication of civil servants with groups of citizens who have special needs (here, it’s important to think not only about people with disabilities but also about other groups like migrants who may not speak local languages);
- Transformation of the decision-making process;
- Transformation of service delivery.
How is it possible to support groups with special needs with the help of AI?
The Flemish government doesn’t want to eliminate physical contact from the process of interaction between civil servants and these groups. However, it is looking for new ways to facilitate and enhance this communication.
For example, when a civil servant in the Flemish public employment service is communicating with job seekers, this conversation is transcribed, written down by AI, and then stored. It means that a civil servant can always find this information if necessary.
This AI system can suggest solutions for each particular job seeker. Even if a job seeker doesn’t speak Dutch, the AI system will translate and transcribe the interview into Dutch for further use.
The idea is not just to have a digital assistant that will help citizens but also a digital assistant for civil servants who are in physical contact with people.
This can make the interaction much more efficient. Moreover, when AI performs multiple backend tasks, civil servants will have more time and force to provide physical and emotional contact to citizens who need it.
Another project of the Flemish public employment service relies on deep learning technology to better match job seekers and job offers. But it is done not only by analyzing people’s formal education or diplomas. The system can also consider competencies that people acquire outside of formal education, for example, thanks to their hobbies.
The growing need for AI translators
For successful AI projects, it is required to have enough data that has value for each particular use case. But it is also crucial to have an ethical awareness and fully understand what can be done with this data and what shouldn’t be done.
That’s why Hans highlighted the importance of having so-called AI translators.
Working for the government, these specialists should be capable of translating a business need as it is formulated by a server servant into a technical solution. That’s why knowledge from both the administrative and technical worlds is a must.
Advice for public sector leaders
According to Hans, at this phase of AI evolution, the key thing that public sector leaders should do is to start experimenting with AI.
He said that there are always civil servants who are ahead of the rest. They are early adapters. And for public actor leaders, it would be helpful to cooperate with such people. These experts can provide oversight of what is being done and whether new initiatives respect the regulations.
In the Flemish government, this role belongs to the AI expertise center.
As you see, every country has its own way of innovation and digital transformation, which is explained by the peculiarities of local society and administration. However, there is one thing that they have in common. That’s their readiness to change their approaches and adapt to the new reality.
Want to learn more about the experience of different countries and businesses in experimenting with AI? Don’t miss the next episodes of the Innovantage podcast! Stay tuned!
Read more:
AI in Education: How LLMs and Oxford AI Research Are Transforming Education in 2025
Neural Networks & Expert Systems: the AI Revolution in the Era of Digital Transformation
Deep-Tech Startups and AI Innovation: Insights from Leesa Soulodre on the Innovantage Podcast
AI ecosystem in Lithuania: How the country is leading innovation
Building an AI-Powered Government: Estonia’s Approach to Data and AI Innovation
Dual-Use Technology and AI Innovation: Their Role and Development
How AI in Retail Is Transforming the Industry: Insights, Trends, and Real-World Examples
Emerging Technologies in 2025: AI, Data Science, and the Future of Innovation
