Modern cities are increasingly moving beyond simply collecting data about urban conditions to taking active measures based on a deep understanding of the situation. Artificial intelligence (AI) is becoming the key tool in this transformation, helping cities evolve into ‘cognitive’ entities. This perspective is highlighted in a piece by Renu Navale, Vice President at Intel and General Manager of the Health and Life Sciences Group.
Imagine this scenario: 60,000 fans are leaving a stadium. Cameras might detect a crowd surge near one exit, but observation alone won’t solve the problem. A decision needs to be made: open an additional exit, reroute the flow of people, or adjust staffing – and it needs to happen in real-time. Similar challenges face intersections, hospitals, and retail spaces: people are constantly moving, conditions change, safety is paramount, and operational processes must continue uninterrupted.
For years, the concept of a ‘smart’ city or space primarily focused on connectivity. Sensors captured events, cameras provided visualization, and dashboards presented data. AI, however, opens up new possibilities by helping cities understand context, determine the necessary next steps, and translate that knowledge into concrete actions. This is the essence of the transition from a ‘smart’ to a ‘cognitive’ city.
Cognitive Systems Unite Observation with Action
The evolution of this process can be described by a simple cycle: “Sense. Understand. Decide. Act. Learn.”
A connected system might inform an operator about traffic congestion. A cognitive system, however, analyzes these conditions within their context and helps determine the most appropriate response. This principle applies to utilities reacting to pressure drops, hospitals coordinating resources, or stadiums managing crowd flow.
Intel identifies three complementary layers that facilitate this transition:
- Edge AI: Processing data directly where it is generated.
- Agentic AI: Transforming gathered information into decisions.
- Physical AI: Implementing these decisions through machines and infrastructure.
Crucially, this doesn’t mean removing humans from the process. Instead, it’s about empowering people to manage increasingly complex environments, while retaining human oversight where it’s critically important.
Cities Are Already Beginning the Transition
Many urban areas have yet to fully embrace this shift. According to a ThoughtLab study, “Building a People-First, AI-Powered City,” only 17% of cities meet the criteria for ‘AI Leaders.’ These leaders demonstrate high levels of adoption for advanced AI technologies like agentic AI and physical AI, as well as digital twins, compared to other cities.
The study also projects that the use of agentic AI among ‘AI Leaders’ will grow from 42% to 84% within three years, while the application of physical AI will increase from 47% to 79%. This data points to a widening gap between cities that are beginning to combine analytics with action and those that are still building foundational capabilities.
The location of computation is also a significant factor. In Sarasota, Florida, computer vision, utilizing existing traffic cameras, was integrated with a traffic management system. Intel and Derq report a 33% reduction in accidents, including a remarkable 90% reduction at one specific intersection. While a specific example, it illustrates a trend: local event detection combined with existing systems for timely action, while cloud services continue to train models and analyze information from various sources.
Leverage Existing Urban Systems
Several real-world barriers hinder broader adoption. The ThoughtLab research indicates that 84% of cities are concerned about unexpected costs, 82% lack AI expertise, 81% struggle to prove ROI, and 76% face limitations due to siloed data and duplicated solutions.
This list can be viewed as a product specification: low total cost of ownership, solutions that integrate rather than create new ‘islands,’ tools accessible to existing staff, and a clear path to demonstrating results. Cities have already invested in cameras, sensors, and operational systems. The next step should build upon these investments, not automatically replace them.
Create a Shared Foundation for Future Innovation
Cities cannot treat every AI opportunity as a separate technology stack. Mobility, safety, utilities, healthcare, sustainability, and public services all have distinct tasks but increasingly require common foundational capabilities: local intelligence, trustworthy data, security, management, and interoperable infrastructure. If each use case leads to the creation of a new technological ‘island,’ AI risks exacerbating fragmentation.
This is the foundation of Intel’s approach to ‘cognitive cities’: a shared Edge AI foundation designed to be:
- Open, to preserve choice.
- Local, to meet governance requirements.
- Efficient, to optimize power and cost.
- Proven, with a path from pilot projects to deployment.
Intel offers a platform spanning silicon, systems, and software, including the Open Edge Platform and Metro AI Suite. This enables cities to scale Edge AI, agentic AI, and physical AI capabilities, while partners can develop their own services. Intel’s role is not to build the city itself, but to provide the foundation upon which others can build and differentiate their solutions. Intel’s broad, open ecosystem approach gives cities a wide choice of vendors to achieve their goals quickly.
Keep People at the Center
As AI’s influence in the physical world grows, trust becomes even more critical. The ThoughtLab study reveals that 78% of cities cite excessive surveillance and loss of privacy as a barrier. It also found that 79% of ‘AI Leaders’ have formal transparency policies, compared to 34% of other cities. Furthermore, 68% of leaders limit data collection and retention, versus 49% of other cities.
These findings position protective mechanisms as features, not afterthoughts. Cognitive systems require clear authorization levels, audit trails, local control, and human involvement in critical decision-making scenarios.
A ‘cognitive city’ should not act simply because technology makes action possible. It should help people make better decisions, automate appropriate tasks, and retain human judgment where it matters most.
Start with One Place and One Meaningful Problem
Cities don’t need to wait for a fully autonomous future. They can begin with a single space, a single use case, and a single partner, built on a foundation that allows subsequent scenarios to evolve without starting from scratch.
The promise of a ‘cognitive city’ is not a city that runs itself. It is a city that helps people see sooner, decide better, and act effectively when it truly matters.
Learn more at intel.com/cognitivecities.
