A human-centred promise, a structural blind spot - Why we should rethink Amsterdam’s AI vision through the Lens of the Commons

Amsterdam’s Vision on AI does not read like a tech manifesto. There is no language about disruption, no promise that algorithms will finally untangle the city’s complexity. Instead, the tone is cautious. The municipality speaks of dignity, justice and sustainability. It avoids calling AI intelligent and prefers the term “intensive automation.” The emphasis is not on brilliance, but on infrastructure,on systems that process data, consume energy and reshape administrative work.

Data reflects inequality. Algorithms reproduce bias. AI systems depend on energy-intensive infrastructures and global supply chains. Automation can weaken human judgement in public services. The city does not pretend that technology is neutral. 

But honesty alone is not a political position. It becomes meaningful only when placed next to the city’s broader ambitions.

In the past few years, Amsterdam has presented itself as willing to rethink its economic foundations. In 2020, during the early days of the pandemic, it adopted Kate Raworth’s doughnut model as a framework for recovery. The idea was simple but ambitious: the economy should guarantee a social foundation while staying within ecological limits. The inner ring represents the minimum conditions for a dignified life, such as access to housing, healthcare, education, income and political participation. No one should fall below that threshold, and economic activity should not overshoot planetary boundaries.

The doughnut was framed as more than a sustainability tool. It implied a break with extractive growth and a willingness to question how value is produced and distributed. Who benefits from economic activity? Who absorbs its costs? Who controls the infrastructures that shape everyday life?

AI systems used by the municipality rely on infrastructures it does not own. Cloud services, computational power and advanced models are controlled by large technology companies. The city pays for access. Public data is processed within systems governed elsewhere. Amsterdam can regulate how it uses these tools, but it does not control the backbone.

The Vision on AI recognises dependency on major tech firms as a risk. It mentions concentration of power and expresses support for open-source solutions – where possible. Yet it’s all framed from the perspective of a responsible user. The focus lies on transparency, oversight and ethical safeguards.

This is where the tension with Amsterdam’s ambition to empower the commons becomes stronger.
The city has actively engaged with the idea of urban commons. Research and policy initiatives have examined how collectively managed resources interact with municipal governance, showing how collaboration can strengthen commons initiatives. They also show how easily collective projects can become integrated into institutional structures, shifting from independent actors to service providers.

Inviting residents to discuss digital policy is not the same as granting influence over digital infrastructure

But participation does not automatically equal shared power. Inviting residents to discuss digital policy is not the same as granting influence over digital infrastructure. Empowering the commons in a digital context would mean addressing ownership, investment and governance. It would mean asking whether data and computational systems should be treated as shared resources rather than services purchased from external providers.

The current vision gestures toward public values, but falls short of structural change. It addresses dependency rather than challenging it. A similar hesitation is displayed when it comes to limits. The document acknowledges that not everything that can be optimised should be; yet, it assumes that AI will gradually expand across municipal functions, maintaining the possibility of deliberately keeping certain domains human-empowered abstract.

Within a doughnut framework, limits are not signs of failure, but conditions for sustainability. If ecological ceilings matter, so do digital ones. Deciding that certain public services should not be automated is not resistance to innovation, it is a political judgement about accountability.

Of course, a municipality cannot single-handedly rebuild the global digital economy. Amsterdam cannot compete with multinational corporations in developing large-scale AI systems. But structural change does not require complete independence; cities can collaborate, invest in open alternatives and strengthen internal expertise, defining red lines in sensitive domains. These choices shape the degree of dependency – and self-affirmation.

Amsterdam’s AI vision is reflective and measured. What remains uncertain is whether that discussion will extend to ownership and power. Ethical use addresses how technology is applied, structural alignment asks who governs it.

If Amsterdam wants its digital policy to reflect its doughnut ambitions and its rhetoric of empowering the commons, it will need to move toward deeper questions of control and collective agency.