AI Doesn’t Know What It Feels Like to Walk Through a Space
By Atakan Guven, Director of Urban Analytics – ERA-co
In recent years, artificial intelligence (AI) has become an indispensable tool for urban analysis, augmenting how we observe, quantify, and project the dynamics of city life. Yet, no matter how precise the algorithm, how comprehensive the dataset, or how refined the model, AI cannot truly capture the authentic feel of a place.
AI doesn’t know what it means to walk down a narrow street lined with fruit vendors, or sit on a bench where generations have gathered, or to navigate the spontaneous choreography of a morning commute. It also lacks proprioception, memory, and lived experience. These are not limitations to be fixed, but reminders of the epistemological boundary between observation and embodiment.
Cities Are Not Optimization Problems
To understand where AI fits (and where it doesn’t), we must first resist the urge to treat cities as problems that can be optimized like code. As Jane Jacobs articulated, cities behave as “organized complexity” – environments where innumerable elements interact simultaneously in non-linear, often unpredictable ways.
AI can parse some of this complexity, identifying patterns in mobility, infrastructure strain, or environmental risk. But meaning, identity, and human connection are not outputs of a model. They emerge through situated practice – people’s movements, choices, rituals, and adaptations over time.
This distinction is crucial. When cities are treated as puzzles to be solved, AI becomes prescriptive. When they are understood as a lived system, AI becomes a tool for observation, reflection, and informed decision-making.
What AI Sees, and What It Misses
AI excels in pattern recognition, video feeds, mobile phone signals, satellite imagery, and street view data, which can reveal how people move through space at different times and under different conditions. These observational tools are powerful.
In a recent proof-of-concept study on visual perception of urban form, AI-driven assessments captured spatial and visual attributes such as enclosure, human scale, and spatial complexity – factors deeply tied to how a place is experienced. These metrics help characterize neighborhoods, inform growth strategies, and even anticipate behavioral responses.
However, AI’s perceptual reach is still limited. It doesn’t know that a narrow alley might feel safe in one city and threatening in another. It cannot differentiate between the ambient buzz of a market and the alienation of a transit hub unless we, as designers and analysts, ask the right questions of the data and apply the right interpretive lens. Otherwise, we risk letting AI read on the surface, ignoring the sociocultural depth that gives a place its meaning.
When The Number Points One Way, And We Go Another
One of the most instructive moments in our recent work came when an AI model recommended a transit intervention based on foot traffic analysis. Although the data was technically sound, our design team, drawing from community consultation and cultural knowledge of the site, identified an underlying reason for the traffic: a gathering place with deep local significance. Altering it for efficiency would have erased its social function.
This kind of counterpoint matters. It reminds us that human judgment, especially the kind rooted in social, cultural, and historical fluency, must remain central in city-making. AI can see movement; people understand meaning.
Precision vs. Presence
There’s an emerging tension between the precision of computational tools and the qualitative experience of urban life. In public space planning, for example, highly granular datasets offer a granular view of behaviors. Yet that same granularity can create an illusion of completeness, masking what isn’t measured: joy, discomfort, nostalgia, tension.
This is where placemaking must resist reduction. A city’s soul is not in its sensor outputs, but in its rhythms – how people linger, how sound bounces between buildings, how festivals reconfigure a plaza, how elders teach youth which streets to walk at sunset. AI can detect footfall, but it cannot interpret footsteps.
Partner, Not Proxy
It’s important to remember that AI is not a proxy for design, but a partner in decision-making. AI enables us to ask more effective questions, test hypotheses, and visualize potential futures. It sharpens our observational lens, but it does not replace dialogue, intuition, or imagination. These remain uniquely human tools, especially when designing for inclusion, authenticity, and cultural rootedness.
Even when confronting global challenges like climate change, migration, or social fragmentation, AI’s role is supportive. For instance, while predictive models can help us map future flood zones or identify infrastructure vulnerabilities, creating urban resilience is about more than anticipation; it’s about fostering adaptability, cohesion, and a sense of belonging. These are social systems, not data systems.
Final Thoughts
AI is a powerful mirror that reflects what we choose to observe, but mirrors are not windows, and they are certainly not pathways. They show us form, not feeling.
In that sense, AI doesn’t know what it feels like to walk through a space. That’s our job. And it always will be.
What matters most is how we choose to integrate these tools into our practice. If we allow AI to dominate decision-making, we risk building cities that are efficient but emotionally sterile. If we instead position AI as a companion to human judgment, one that sharpens our capacity to see patterns, test scenarios, and anticipate risks, we preserve the role of culture, memory, and community in shaping urban life.
The future of cities will not be defined by the accuracy of our algorithms. It will be defined by the wisdom with which we interpret them.
Atakan Guven is the Director of Urban Analytics at ERA-co. At ERA-co, Guven works across multi-disciplinary teams on projects varying from pedestrian circulation and public realm strategies to concept masterplans. As part of the Urban Strategy and Planning team, he identifies and implements evidence-based design insights to make better places, spaces, and systems. ERA-co has been featured in Connected World, High Profile, and more.