Dialogic conducted recent research on algorithms and AI within municipalities. What was the exact question?

Municipalities are increasingly exploring how they can utilise algorithms and AI to analyse issues faster and better and to make more targeted decisions. However, their use can also have an impact on residents and businesses. The rise in the use of algorithms and AI raises questions for the audit office about how municipalities deal with algorithms and AI. Therefore, we researched where algorithms and AI are already being used, what risks are involved, and how municipalities are preparing organizationally and administratively for responsible use.

In your opinion, where is the biggest challenge for municipalities?

The primary challenge lies not in the technology itself, but in the organization around it. Who takes ownership? What knowledge is required? How do you make agreements about responsible use? And perhaps even more importantly: how do you prevent AI from becoming something driven by a few enthusiastic leaders, and how do you ensure that the organization maintains sufficient control?

A significant development is the growing use of AI within municipalities. Especially generative AI is receiving a lot of attention, for instance with systems like ChatGPT and Microsoft Copilot. In practice, you often see enthusiastic leaders who drive a lot of change, alongside teams who prefer to continue working as they always have. This makes the organization vulnerable if knowledge and responsibility lie too much with individuals. Therefore, central control is necessary to ensure that responsible use is not dependent on random initiatives but is more broadly secured.

What does this mean in practical terms for a municipality looking to enhance its AI maturity?

Responsible use of algorithms and AI starts with knowing what is happening, determining where you want to go, and being able to organize that employees, processes, and systems are equipped for it.

• When it comes to knowing, it's about having an overview: which algorithms and AI applications are already in use, and how are the responsibilities divided?
• Next comes wanting: where does the municipality want to use AI, where not, and under what conditions? This requires clear policies and distinct choices.
• Being able to then revolves around practical feasibility: do you have the right people, knowledge, processes, and agreements in place to actually organize this.

This threefold approach - knowing, wanting, and being able to - helps municipalities transition from individual initiatives to conscious and manageable use of algorithms and AI.

What stood out to you personally in the research?

What stood out to me the most is that municipalities often already use more algorithms than they initially think. When people think of AI, they quickly consider generative AI like Copilot or ChatGPT, for tasks such as writing, summarising, or analysing texts. AI policy-making tends to focus mainly on this. However, rule-based algorithms are equally important. These have often been part of municipal processes for a long time and can have a significant impact on residents. There is a risk of limited oversight of the use of these decision rules. Simultaneously, I observed a lot of positive energy within municipalities. Municipalities want to learn, experiment, and establish clear frameworks.

What can municipalities learn from each other?

Municipalities can learn a lot from each other, precisely because they grapple with similar issues. Even a municipality that has limited experience with algorithms will encounter them through joint arrangements and suppliers. Therefore, 'doing nothing' is actually not a neutral choice. Inspiration for deploying (new) algorithms and gaining best practices can be drawn from other municipalities. This way, the municipality stays informed about the latest developments. Larger municipalities, however, have more capacity. Smaller municipalities can position themselves as 'smart followers'.

What does this mean for the upcoming years?

For the upcoming years, this means that municipalities can no longer treat AI as merely an experiment but must increasingly integrate it as a fixed component of their organization. The focus shifts from "what can be done with AI?" to "how do we ensure responsible deployment and central control?". The core remains the same: the ambition for responsible AI use must be embedded in the daily practice of the entire organization.