Baarsma and Nieuwenhuis point out that many city services and policy domains are still stuck in longstanding bottlenecks, from waste collection and maintenance planning to traffic management and infrastructure upkeep. They note that AI offers tools that can combine and analyse large amounts of data, enabling predictions and proactive planning that traditional approaches struggle to deliver.

According to the authors, Amsterdam should not treat AI only through the lens of rules and restrictions. Instead the city should open up public spaces, data environments and governance experiments where startups, researchers and civic innovators can test ideas, measure results and refine solutions in real urban conditions. They see experimentation as essential to building experience, avoiding pitfalls and understanding how AI systems interact with complex social and administrative processes.

The opinion piece highlights that without such room for experimentation, policy will remain stuck at theoretical possibilities and ambitions will fail to translate into practice. The authors advocate for initiatives that allow iterative development, learning from failure and co‑creation with communities, rather than top‑down mandates that overlook local realities.

Baarsma and Nieuwenhuis stress that this approach will help Amsterdam address not only technological challenges, but also broader issues of trust, transparency and public acceptance. By involving citizens and stakeholders early on and by making experiments visible and accountable, the city can ensure that AI solutions are socially robust and widely beneficial.

The piece ultimately frames experimentation as a strategic asset rather than a regulatory risk, arguing that cities that embrace learning through trial will be better positioned to lead in AI innovation and solve persistent urban problems.