Cities invest in AI but lack governance to make it work

Algorithms have improved. Data availability has increased. Interoperability standards exist. What has not kept pace is organizational readiness. City governments remain structured around vertically organized departments, each with its own budget, leadership, performance metrics, and data ownership rules. These structures were not designed for cross-domain intelligence.

Cities across the world have invested heavily in sensors, data platforms, and artificial intelligence tools to support development goals. Despite this surge in deployment, most smart cities remain unable to connect these technologies into coherent, city-wide systems. A new academic study published in Urban Science finds that the failure is not technological, it is organizational.

The study, titled The Organizational Transformation of Artificial Intelligence in Smart Cities: An Urban Artificial Intelligence Governance Maturity Model, argues that fragmented governance structures are the primary reason artificial intelligence has not delivered systemic benefits in urban environments. While AI systems perform well within isolated departments, cities lack the institutional capacity to integrate them across domains.

AI is working, cities are not integrating it

Past analysis by the same authors showed that more than 90 percent of AI systems deployed in the urban built environment operate at the lowest levels of integration. These systems often deliver strong results in narrow domains such as traffic optimization, energy monitoring, or infrastructure maintenance. Performance gains within silos are significant. What is missing is coordination.

According to the study, this disconnect produces what the authors describe as a fragmentation paradox. Artificial intelligence demonstrates high technical effectiveness when deployed in isolation, yet cities fail to achieve broader system-level improvements. Transport systems do not coordinate with energy networks. Emergency services do not share real-time intelligence with environmental monitoring platforms. Urban systems that could be optimized together remain separated by institutional boundaries.

The research makes clear that this fragmentation persists despite rapid advances in AI capability. Algorithms have improved. Data availability has increased. Interoperability standards exist. What has not kept pace is organizational readiness. City governments remain structured around vertically organized departments, each with its own budget, leadership, performance metrics, and data ownership rules. These structures were not designed for cross-domain intelligence.

Most municipal funding models reinforce this separation. Investment flows toward short-term pilot projects with clear, department-specific outcomes. Shared infrastructure, long-term integration, and cross-sector coordination struggle to secure political and financial support. As a result, cities accumulate disconnected AI systems without building the governance capacity needed to link them.

The authors argue that this pattern explains why smart city programs often appear busy but fail to transform how cities function. Without governance reform, additional AI deployment simply deepens fragmentation.

A governance maturity model for urban AI

To address this gap, the study introduces the Urban AI Governance Maturity Model, a framework designed to help cities assess and improve their readiness for integrated artificial intelligence. Unlike existing smart city or IT maturity models, this framework is built specifically around the governance requirements of cross-domain AI systems.

The model defines five stages of maturity. At the lowest level, cities rely on ad hoc, siloed AI projects with little coordination and no shared strategy. This is the dominant condition observed across global smart city initiatives. The next stage reflects growing awareness of integration challenges, where cities begin to inventory data assets and discuss governance but lack consistent implementation.

At the middle stage, formal structures emerge. Cities define city-wide AI strategies, establish shared governance rules, and mandate interoperability standards for new projects. Only at this point does meaningful integration become feasible. Higher stages involve operational city-wide data federation, cross-domain analytics, and finally AI-driven system optimization, where multiple urban systems are coordinated simultaneously.

Notably, the model does not treat technology as the primary driver of progress. Instead, it identifies five organizational capability areas that must develop together. These include strategic leadership and sustained investment, organizational structure and culture, data governance and policy, technical capacity and interoperability, and trust, ethics, and security.

The study stresses that maturity in one area cannot compensate for weakness in another. Advanced analytics cannot overcome unclear leadership. Open standards cannot function without enforced data governance. Ethical frameworks are ineffective if they are not embedded in operational decision-making. Cities that advance unevenly across these areas remain unable to scale AI integration.

Why governance must evolve with technology

The authors introduce the concept of a governance–technology interlock to describe this relationship. Certain forms of AI deployment are only possible when corresponding organizational capacities are in place.

For example, city-wide data federation requires more than technical infrastructure. It depends on shared data ownership rules, standardized data catalogs, formal sharing agreements, and leadership with the authority to enforce them. Without these elements, attempts to integrate data across departments often fail or stall indefinitely.

Similarly, advanced AI-driven optimization across multiple urban systems requires a governance environment capable of managing ethical risk, accountability, and security at scale. The study warns that deploying such systems prematurely can undermine public trust and increase institutional resistance to future innovation.

The paper shows these dynamics through comparative analysis of well-known smart city cases. High-performing cities demonstrate strong central leadership, mandated interoperability standards, and dedicated governance bodies with cross-departmental authority. These conditions enable integration beyond pilot projects. Cities with weaker governance structures show progress in individual domains but struggle to scale coordination.

The authors are careful to note that governance reform is not a purely top-down exercise. Organizational culture, incentives, and performance metrics play a critical role. Departments accustomed to optimizing their own outcomes must be rewarded for collaboration rather than penalized for sharing control. Without cultural change, formal governance structures remain ineffective.

The study also highlights the role of ethics and security as core governance concerns rather than afterthoughts. As AI systems increasingly influence urban decision-making, cities must address transparency, accountability, and cybersecurity at the same pace as technical deployment. Failing to do so risks public backlash and regulatory intervention that can stall innovation altogether.

Implications for smart city policy

The study suggests that continued investment in AI without corresponding governance reform is unlikely to produce meaningful returns. Cities may accumulate technology assets while remaining structurally incapable of using them together.

The authors argue that the first priority for cities should be establishing clear, city-wide AI strategies supported by sustainable funding models. This includes investing in shared digital infrastructure rather than isolated projects. Leadership roles with cross-departmental authority are essential, particularly in data governance.

Procurement policies also emerge as a critical lever. Mandating open standards and interoperability requirements can prevent vendor lock-in and reduce technical barriers to integration. However, these mandates must be enforced through governance structures that extend beyond individual departments.

The study reframes ethical AI governance as a prerequisite for scale rather than a constraint. Robust oversight mechanisms, clear accountability, and proactive security practices are necessary conditions for deploying AI systems that affect multiple urban domains.

Perhaps most importantly, the research challenges the prevailing narrative of smart city development. It suggests that the next phase of urban innovation will be defined less by new technologies and more by institutional transformation. Cities that fail to adapt their governance structures risk falling behind, regardless of how advanced their technical tools become.