LAVAi marks a coming-of-age moment for an architectural discipline entering the AI era with urgency and uncertainty. How are you approaching this paradigm shift?

Tobias Wallisser (TW): LAVAi will bring us closer to the future we are beginning to see on the horizon. AI may be a black box, but powerful tools have always forced architects to rethink their processes. The challenge is to unbox AI with curiosity and rigor, and the belief that its inventive promise lies in its capacity to expand how architects interpret complexity, communicate insight, and navigate the unknown.

Let’s begin that navigation with chess. You both like to return to the anecdote about IBM’s Deep Blue computer defeating then-reigning world chess champion, Garry Kasparov, at chess. Why does that moment matter to you?

TW: Because of what comes after it. A computer can defeat the best human chess player at chess, but it can only play chess. A human can play chess, ride a bike, cook spaghetti, and improvise when conditions change. In the chess match between human and AI, the real question wasn’t how strong the AI was; it was how limited. How do you train a system to move across domains instead of optimizing for only one?

Jan Kováříček (JK): Humans follow rules until they fail. AI can step sideways and find strategies we wouldn’t consider. That’s where play comes in. We are looking at multimodal systems where AI is trained on rule-breaking and experimenting with different kinds of data simultaneously. This approach is incredibly relevant for architecture and urbanism, where many of our inherited assumptions about how cities grow, how people move, or what a space is supposed to do are rooted in the limitations of the tools we’ve historically had. We didn’t necessarily believe cities were simple or predictable, but we could only model them that way. What AI allows us to do now is finally engage with the true complexity of urban systems, which are dynamic, adaptive, and often unpredictable—and design in a way that reflects that reality.

Working with AI can be surprisingly playful. We tend to assume its value lies in making things faster or easier. But what makes it interesting is the unexpected variables it brings to the thinking process.

TW: Exactly. Efficiency is the least interesting part. The opportunity we have now is to redefine the game the way we did when digital tools entered architecture and computation transformed analog processes. Architecture is about dealing with uncertainty. AI doesn’t remove that; it makes it visible so that objects can become responsive, and the discipline can shift from one that is defined by static artifacts to one that co-creates organisms.

Does AI turn your projects into living architecture?

TW: Yes. Architects now design hardware and software together—the physical building and the digital systems that shape it. The challenge, though, is that building is slow. Ideas can take a decade to materialize. Cultural corrections take even longer, and resistance to change is still strong, even toward tools that have been part of our practice for years.

AI image generation has already altered how projects take form. Designers can produce highly realistic visual models from ideas that are still in progress. Traditionally, the image came last. Now it comes first. BIM initiated a comparable shift by binding geometry to information, and performance analysis extended that logic further. The digital twin established performance as an active condition within the project rather than an external assessment. All of this transforms the horizon of the discipline. With AI, we can now move from the notion of static singular buildings to that of the city as a living, evolving entity that adjusts over time.

One example lies in how a mix of climate and social conditions is experienced in real time. Weather data can be read alongside platform APIs such as Instagram, X, or Google Maps. Geolocated posts reveal patterns of presence and moments of heightened activity that register the lived experience of social interaction in the city. When information about human movement, communication, and sensation is considered alongside environmental metrics such as precipitation, runoff, or water absorption, new relationships become visible. These intersections open new lines of inquiry into urban systems that we are only beginning to understand and articulate.

This is like the creation of a hologram of a city, or an area of a city, that sees more than the obvious physical elements. I imagine this view expands when you mix AI trained on your own data with collaborations from other fields, like biologists, gamers, psychologists, and climate scientists.

TW: Absolutely. LAVA has always worked in this cross-disciplinary way. We experiment in as many ways as possible before translating ideas into practice.

JK: Climate modeling is empowered by this way of working. Many of us are familiar with the IPCC assessments and the fact that current developments fall short of their targets. In our work on Future Cities, the key shift is in moving beyond reliance on historical climate data toward constructing future-oriented scenarios. We use AI to generate synthetic datasets based on projected climate conditions, and adjust these to existing weather models to reflect possible futures. This new, more complex and dynamic data is then fed back into our simulations. In this way, instead of designing for the past, we begin to design for what is likely to come.

TW: A broader question becomes whether we can apply a comparable method to human-made systems such as mobility networks, housing markets, energy grids, and patterns of land use. Climate science models the interaction of variables over time; urban planning often assumes stable inputs and fixed end states. If we simulate how these urban systems evolve in relation to one another, planning fundamentally transforms. The focus moves from optimizing for a single projected scenario to preparing for a range of possible scenarios, conditions, and spatial configurations over time.

JK: Machine learning is extremely good at pattern recognition across scales. Cities become ecosystems, spatial arrangements, living systems.

TW: That’s why I don’t like the term “smart city.” It implies optimization for efficiency or profit. We’ve been calling one of our projects the “Conscious City”. Not smart, but aware, and reprogrammable as a city with senses that are capable of interacting and responding with its many varied occupants.

This novel new architectural gaze also means questioning assumptions about what cities are, what happens in them, and how they behave.

TW: Yes, and in this gaze, assumptions like cars aren’t even a given. Cities have existed without them. AI allows us to reconsider mobility in order to emphasize access, connectivity, and ecology instead of solely focusing on traffic flow. We can be radical with our visioning, given these new tools. By engaging AI in the discussion of speculative questions, the framework begins to shift, and the exchanges expand the range of possibilities we can examine with analytical depth.

With these tools, you might also include the emotional element in design that you’re interested in at LAVA, such as creating a feeling in a room, a space, an ecosystem, or a city. Imagine giving a city a vibe.

TW: We’re working on that. In large-scale projects like Expo 2030, there are areas designed for rest after moving through high-stimulation environments with a great deal of input. These spaces need to radiate calm, even when they sit at the center of intense activity. That is a significant challenge. With co-pilot assistance, we can begin to work toward these less tangible goals and test how spatial and environmental parameters influence perception and mood.

Can you talk about how this works?

TW: It starts with the senses. Smell is very important. In exhibitions, for example, our orientation depends largely on sight, occasionally on touch. In spaces intended for rest, smell and sound establish an atmosphere. A certain scent or acoustic quality can evoke the impression of a city or a village and guide our movement through intuition. When the atmosphere resonates, we linger; when it does not, we move on. Much of urban navigation unfolds in this quiet, sensory register. AI can construct speculative scenarios around these subtle conditions and analyze how they shape patterns of movement, perception, and memory.

Urbanism rarely plans at this level of experience. Master plans define buildings and retail zones. Shop owners introduce displays and goods, and their associated smells, colors, and textures. These subtle sensual qualities could enter the design brief from the outset, so that space carries atmosphere by intention, not by accident.

Are you defining a new design genre here? Could we call it “sensual urbanism”?

TW: I think so, yes. Imagine designing cities like stories, using AI to shape experience, well-being, and narrative. Buildings and entire urban ecologies that are predicted, sensed, and choreographed. I’m thinking of Disney’s Imagineering, a portmanteau for the hybrid combination of imagination and engineering—but amplified, reprogrammable, and adjustable over time. AI imagineering for the next era of city-making, where every street, sound, and scent plays a part in the design of life in that city.

I’m envisioning design with birds and insects, water flows, infrastructure, what smells good, what memories can be triggered, and what makes people feel good in a city. Is this what you mean?

JK: Yes. For an architect, some factors, like shade, green space, permeability, and surface type, are obvious. But solving these across an entire city is a different challenge, and AI co-piloting is becoming indispensable in this. In our Future City research, we tested the concept with Berlin, which, like many European cities, has a well-managed geoportal full of urban data. Without AI, interpreting each district manually would be impossible. But we were able to “imagineer” them by digitally layering heat maps with green versus built areas, asphalt coverage, biodiversity, and human activity. AI uncovered patterns and insights here that guided us into seeing a more complex and harmonious urban flow.

TW: In this work, we’re most interested in looking at relationships, not at a single design parameter. A lot of people treat CO₂ as the new currency and focus only on emissions. It matters, but so do a lot of other factors. Future thinking is about experimenting with as many variables as possible and working with them speculatively until a considered solution is arrived at that manifests the sustainable, symbiotic positivism we are designing the future to meet.

JK: We are facing conditions that exceed the scale of inherited human consciousness. Addressing climate change may require forms of synthetic intelligence trained on patterns far broader than any previous system could process, systems that integrate nature’s intelligence alongside our own. When we embed migratory routes, ecosystems, and habitat behaviors into our data sets, fauna and flora begin to register as agents rather than as background. What this presents is the possibility of symbiotic intelligence, and a mode of thought where human and other life-form systems are modeled, sensed, and negotiated together. At LAVAi, we are moving toward that horizon of symbiotic intelligence.

Working with AI is facilitating an unprecedented leap in architectural possibilities. It’s so new I even want to call it an emerging “architectural consciousness”.

JK: We’re seeing exceptional learning potential, where humans are sometimes guided by technology in new ways. For example, a junior architect at our office was analyzing a complex geometrical surface. The simulation models were working, but interpreting the results was challenging. Using an AI co-pilot, which had been trained to explain structural analysis outputs in clear terms, he was able to translate the raw numbers into actionable insight. This opened up a dialogue between the architect and the model that enabled more informed decisions to take place in the next iteration by enhancing understanding without replacing expert judgment.

TW: Architecture already extends beyond the physical. We shape spaces within a continuum. Working with this technology allows that to become more visible and more present in our complex negotiations with the living and built environments. What excites me most in this new architectural playing field is the playfulness, where prompts focus on qualities and dialogue replaces determinism. German architect Günter Behnisch once said that unbuilt ideas land on the compost heap—but they fertilize the next ones.

JK: In AI, we call that the “latent space”, a compressed representation of possibilities, where complex data is encoded into patterns that can be sampled and combined in new ways. It’s how AI helps reveal options that remain hidden in traditional workflows.

TW: Which connects to the Eastern idea of the void, where emptiness becomes infinite potential. AI operates in the same register and pushes us to think multi-dimensionally, with long-term visions around growth, adaptation, and learning.

JK: The challenge now is creating a sustaining symbiotic best practice of architecture based on a completely new vision of our discipline.

TW: One that includes such new concepts as sentient buildings, conscious cities, symbiotic and sensual urbanism, and notions of AI imagineering with care inside the rise of a new architectural consciousness.