MIT's Carlo Ratti Sees Seoul as a Living Laboratory for Cities That Learn

Seoul offers a place to test urban change at scale

Seoul's combination of density, digital connectivity and residents who quickly adopt new services makes it an unusually promising place to test the next generation of urban infrastructure, according to Carlo Ratti, the Italian born architect and researcher who leads the MIT Senseable City Lab. Speaking at the Smart City Expo World Congress in Barcelona, Ratti pointed to South Korea as a place where ideas can be tried across a large urban population and improved quickly.

Ratti's argument is that cities should be renewed continuously through data and design, rather than built once and left to age. Streets, buildings and utilities could sense changing conditions, adjust their operations and evolve alongside residents. He describes this direction as learning infrastructures: physical systems that use information to respond to what is happening around them.

South Korea brings together digitally connected citizens, public agencies experienced in operating infrastructure to demanding standards, and a strong base of electronics makers, software companies and startups. Seoul also has high public transport use and near universal smartphone access. These conditions give urban experiments both a large testing ground and opportunities for feedback.

Population decline adds urgency to that approach. Korea, Italy, Japan and China face shrinking populations in the coming decades, leaving city leaders to provide comfort, safety and convenience with fewer workers and tighter budgets. Ratti's case favors upgrading existing neighborhoods and networks instead of building new districts from scratch.

Technology is only part of the challenge. Sensors, fiber optics and connected controls need rules governing ownership, access and decisions. For Seoul, the central task is to make experimentation useful to residents while giving them confidence in how their information and shared public spaces are managed.

What makes infrastructure capable of learning?

Traditional infrastructure is built, commissioned and maintained, with operating settings recorded in manuals and major upgrades arriving decades apart. Learning infrastructure adds a continuous feedback loop. It measures activity and conditions, analyzes the information to anticipate what comes next, and adjusts operations as circumstances change.

Traffic signals could change their timing in minutes rather than months. Building ventilation could respond to the number of people inside and the quality of outdoor air. Drainage systems could open and close to help absorb sudden storms. The purpose is responsiveness, with equipment adapting to actual demand rather than following a fixed setting.

Fiber and power lines beneath streets support connections to cameras, acoustic sensors, air quality monitors and traffic detectors. At street level, lidar, Bluetooth and beacons from connected vehicles add information. Lidar uses light to measure surroundings. Inside buildings, smart meters, thermostats and lighting networks provide detailed feedback about energy use and conditions.

Open standards, which allow different systems to exchange information, can help organize these networks into a useful whole. Potential benefits include faster bus journeys, lower energy bills in schools, safer pedestrian crossings and fewer flooded basements during intense rain.

Weather can damage or disrupt sensors, and the information they gather can be incomplete or difficult to interpret. Cybersecurity needs attention from the beginning. Connecting equipment and software across agencies and vendors is also difficult. Cities must decide how much information they need while protecting privacy. Ratti sees Seoul's scale, skills and civic energy as strengths for working through these problems with residents at the center.

Why Seoul's residents matter as much as its networks

Seoul's metropolitan scale gives new services a large population to serve, while its digital connections allow residents to interact with the city through mobile applications and kiosks. Those everyday encounters create opportunities to learn whether a service works and what needs changing.

Electronics manufacturers, software firms and startups provide another advantage. Their presence can help urban projects move from an idea to deployment without long supply chains. Combined with a public accustomed to trying digital services, this creates conditions for rapid testing and revision.

Accessibility, affordability and privacy still need to shape each project from its first stage. An experiment that works only in wealthy districts, or requires the newest device, would overlook people who could benefit from it. Seoul's public transport and digital payment services already reach a broad public, providing a foundation for more inclusive projects.

Ratti's vision depends on residents being participants in change. Feedback needs to influence design, including when it challenges the original plan. Fast adoption is useful when a city also checks whose needs a service meets and whose needs it misses.

Who controls the equipment and the data?

Smart city technology can travel across borders more easily than the arrangements governing it. A sensor attached to a street light brings practical questions about who owns it, who stores its data, how long that information is kept and how residents can challenge an automated decision affecting them.

A single data feed may pass through several organizations. A vendor might collect it, a cloud provider process it and a municipal dashboard display the results. Each step creates rights and responsibilities. Contracts that leave these arrangements unclear can cost a city control over access, formats and the timing of information delivery.

Open data schemas give information a shared structure. Data portability allows a city to move information between systems. Short retention periods for raw personal data and independent audits can help protect residents and check whether contractual promises are being followed.

Public technical teams also need the skills to assess the platforms they buy. Procurement can favor modular hardware and open interfaces, allowing individual components to be replaced without discarding an entire system. That matters when a sensor, software service or supplier changes.

Experience across cities places many of the largest obstacles in social and institutional arrangements rather than the machinery itself. Clear answers about ownership and responsibility can help projects expand. Unresolved questions can leave trials stalled or trigger public resistance.

Transparency needs to be visible to residents

Ratti advocates radical transparency: residents should be able to understand what information is collected, why it is needed, how it is protected and when it will be deleted. A public data inventory, clear retention schedules and explanations in ordinary language would make those decisions easier to examine.

Tools inspired by Europe's General Data Protection Regulation can give people rights to access, correct and delete personal information. Independent oversight boards and privacy impact assessments can check whether systems remain consistent with public values. Visible safeguards can help residents feel more comfortable with sensors in shared spaces and with contributing information for public benefit.

Choice is another part of trust. Some services can operate through voluntary participation. Clear signs in areas containing sensors and straightforward complaint channels can reduce uncertainty about what a city is doing.

Algorithm registers can identify the automated systems used in public services. Cities can also allow outside review of models and organize red team exercises, in which reviewers deliberately look for weaknesses before they cause harm. These measures concern public accountability as much as technical performance, and would help determine whether connected systems retain support over time.

Using AI to understand greenery and wellbeing

Artificial intelligence can help planners examine urban environments on a large scale. It can map tree canopy, shade and pavement, estimate movement speeds on sidewalks and identify places where vehicles and pedestrians come dangerously close.

Ratti and collaborators are applying such methods in Korea through Nature That Heals, Seoul. The project seeks to connect people's experiences in parks and streets with measurable features of greenery and design. Its central idea is to learn which surroundings support comfort and wellbeing, then use that information to guide investment.

A shaded bench near a fountain illustrates the approach. If people linger there and experience lower stress, researchers want a system capable of detecting that pattern and informing the creation of more places with similar benefits. The example describes the project's intended use of evidence rather than a completed finding.

Computer vision, software that recognizes features in images, can classify trees, grass, water, shade structures and seating. Combined with sensors measuring local climate conditions and anonymized movement information, it can help identify comfortable summer paths, noisy plazas and places where small design changes could encourage daily physical activity.

The intended result is a map of opportunities rather than a single score for an entire area. Planners could compare conditions block by block, then test whether new trees, a misting station or a sound barrier improves comfort and attracts use.

A wellness project also needs privacy protections

Images and movement information require careful handling. Systems studying public space should avoid identifying individuals, and sensitive footage should be processed at the edge, meaning on the device itself, so that raw video does not need to leave it.

Aggregation combines individual observations into broader patterns. Noise injection adds statistical variation to information to help protect privacy while retaining useful trends. Both approaches can support analysis of how spaces are used without making personal movements the focus.

Teams also need to check for bias. If images from some districts are poorly represented in the material used to train an AI system, those neighborhoods may receive less useful analysis. A project intended to improve wellbeing needs to account for the city beyond its most visible areas.

Nature That Heals, Seoul places human experience alongside measurable physical features. Its value would lie in helping planners test changes and direct resources toward places where comfort and use can improve.

What China and planned new towns can teach Seoul

Ratti uses the term senseable city to describe an approach in which sensors and software help people understand and shape their surroundings. Human needs, rather than a checklist of equipment, guide the design.

Shenzhen offers lessons about speed and scale. Rapid urbanization, manufacturing capacity and an entrepreneurial culture helped make it a workshop for urban technology. International teams study how its public spaces, transport and industry have developed together.

Early showcase projects in several countries, including towns built from scratch, also offer a warning about rigid planning. Neighborhoods can feel sterile when decisions come only from the top and plans leave little room for change. Technology then becomes something added to an empty setting rather than part of an active community.

Mature cities can absorb technology through gradual changes to individual blocks. Active ground floors and a mixture of uses provide everyday activity around which services can develop. For Seoul, the lesson is to test ideas on real streets and in busy stations, and to accept feedback that changes the experiment.

Private businesses remain essential partners. Cities also need to protect themselves against dependence on a single vendor and against standardized systems that erase local character. Diverse approaches and open discussion can help connected services improve public life while leaving neighborhoods recognizably their own.

Street lights reveal both the promise and the difficulty

Street lights show how ordinary infrastructure could take on new functions. Replacing existing lamps with energy efficient LEDs can save money immediately. Connectivity can allow lights to dim when nobody is nearby and report faults without waiting for someone to notice an outage.

A connected pole can also carry sensors for air pollution, traffic speed and parking availability. It could support emergency services, help adjust timing for pedestrians and host 5G small cells, which are compact pieces of mobile network equipment. One fixture could therefore serve lighting, transport, communications and public health needs.

Municipal departments usually have separate responsibilities. Lighting crews focus on illumination, transport teams on signals and health departments on air quality. A pole serving all three requires clear ownership of tasks and budgets so that responsibilities are not left unattended.

Modular procurement would allow a city to add or replace sensor cartridges without replacing the entire light fixture. Privacy rules also need to be explicit. Where cameras are used, removing identifying details on the device and keeping footage briefly should be the default. Many cities limit collection to counts and speeds rather than identities, retaining information useful for street design while reducing privacy risks.

Discussions with city officials and industry reveal competing priorities. Vendors often promote proprietary data platforms. Utilities concentrate on reliable lighting. City managers need services that work today, while advocates of open data want systems that can support future applications.

Policy needs to guide purchasing decisions. Hardware should fit the city's decisions about data and public service. Early involvement by residents and protection of their data rights could help connected street lights become a trusted foundation for urban intelligence.

A smaller and older population changes the priorities

Population decline shifts planning toward getting more value from infrastructure already in place. Instead of concentrating on expanding supply, cities need ways to adapt streets, buildings, water networks and energy systems to changing demand.

Adaptive traffic lanes could serve freight loading, buses or cycling at different times of day. Unused offices could become libraries, clinics or rooms for shared study. Water and energy systems could adjust their operations as demand falls, reducing costs while maintaining reliability.

Learning infrastructure offers a way to make these adjustments without long delays. Data about actual use could help services respond at an appropriate scale, supporting the needs of residents as populations and patterns of activity change.

An aging population brings additional priorities. Shorter pedestrian crossings, more benches and better shade can help older people remain active. The networks supporting energy and transport could also carry remote health services and social care.

The strongest projects in this setting would ease daily work for caregivers, help emergency responders arrive faster and make extreme heat less stressful. Comfort and dignity are the intended outcomes. Connected equipment is a means of reaching them.

What progress would look like on Seoul's streets

Bus passengers could experience progress through shorter journeys, with signals detecting approaching buses and coordinating green lights. Shops in areas prone to flooding could face fewer closures if drains were monitored and controlled before sudden heavy rain arrived.

Summer walks could become cooler as the city places shade along routes people already use and checks whether new trees produce the expected temperature reduction. Utilities could use demand information to schedule maintenance, making outages less frequent and shorter. These are examples of the results Ratti's approach seeks.

Public accountability would be another measure of progress. A website could list every sensor type, its location by block, its purpose, the information collected and the period for which it is kept. Registers could explain how algorithms used for parking prices or building inspections are trained and checked for bias.

Residents could download personal information held about them, correct mistakes and request deletion. An independent board could check compliance and publish reports every quarter. These arrangements would make oversight part of the service rather than leaving it separate from daily operations.

Open interfaces could allow startups to build navigation services for people with low vision or delivery routing that reduces nighttime noise. Universities and civic groups could use public datasets to assess climate resilience and public health. Neighborhoods receiving less attention could get early improvements alongside central districts.

Ratti's vision joins these everyday gains with continuous revision. Seoul's role as a living laboratory would depend on experiments becoming useful public services, with benefits reaching a broad range of residents and clear rules governing the systems behind them.

The Bottom Line

• Carlo Ratti, leader of the MIT Senseable City Lab, identifies Seoul as a place to test urban technology at scale and improve it quickly.
• Learning infrastructures use data to adjust streets, buildings and utilities as conditions change.
• Seoul's density, connectivity, technology businesses and public adoption of digital services support experimentation.
• Ownership, privacy, open interfaces and public oversight are central to whether projects can expand and retain trust.
• Nature That Heals, Seoul seeks to connect greenery and urban design with people's comfort and wellbeing.
• Lessons from Shenzhen and planned new towns favor experiments rooted in everyday life and responsive to residents.
• Connected street lights could support several services, but need clear departmental responsibilities and data protections.
• Population decline and aging favor upgrades to existing infrastructure and services that support comfort, care and dignity.
• Success would combine practical gains, such as faster buses and cooler walks, with transparent data management and benefits across neighborhoods.