Installation view of Living Architecture: Gehry at the Guggenheim Museum Bilbao © Refik Anadol, Bilbao 2025 What does it mean for architecture to breathe, shift, and evolve like a living organism? Can buildings dream? These are the kinds of questions that drive Refik Anadol's artistic investigations, where artificial intelligence becomes a medium for reimagining space and perception. As a pioneer at the intersection of AI, data visualization, and immersive art, Anadol has spent years transforming vast datasets into fluid, dynamic environments that challenge our understanding of both physical and digital realities.
In Living Architecture: Gehry, recently unveiled at the Guggenheim Museum Bilbao, Anadol presents a large-scale AI-driven installation as part of the museum's debut in situ series—a program dedicated to site-specific works that engage with the museum's architecture. Taking inspiration from Frank Gehry's iconic forms, the work translates 35 million natural and architectural ethically collected data into an ever-changing spectacle of movement and light. The piece transforms a museum gallery into a fluid, sensory experience, continuously shifting as if the walls themselves are alive. This installation debuts the Large Architecture Model (LAM), an innovative AI model developed by Refik Anadol Studio.
A year in the making, the installation goes beyond aesthetics, exploring how technology shapes our perception of the world. For Anadol, AI is not just a tool but a collaborator—one that enables him to explore questions of perception, memory, and the fluid nature of space. As he explained during the exhibition opening, his interest in AI is not about "techno-fetishism" but about uncovering the invisible systems shaping our world. Going beyond traditional artistic categories, Anadol proposes what he calls "generated reality," envisioning a reality in flux, where AI and data create an environment that constantly shifts, responds, and evolves.
Stepping into the space, visitors find themselves surrounded by a constantly morphing visual landscape. Waves of color and motion wash over the surfaces as if the museum itself is breathing. At times, the imagery evokes natural forces—cloud formations, shifting tides, the slow erosion of landscapes—while at others, it feels purely digital, a world composed of light and code. The scale of the installation is both overwhelming and intimate, drawing the viewer into a sensory experience that feels both futuristic and deeply elemental.
Unlike traditional digital projections, which operate on fixed loops, Living Architecture: Gehry unfolds in real-time. Rather than following a predetermined sequence, the installation functions in what Anadol calls "chapters"—distinct yet interwoven visual states that emerge unpredictably. The work, featuring all-encompassing visualizations of digitized memories created by employing neural networks, unfolds across six interconnected chapters, each representing a stage in the evolution of data into architectural imagination. The installation begins with a vast compilation of data—photographs, plans, and other materials—which is processed to create intricate patterns and immersive environments. As the AI processes the data, it shifts from constructing architectural models to generating abstract, dream-like moments, ultimately returning to architecture and "dreaming" Gehry-inspired forms—fluid and speculative in nature.
"We have an archive of possibilities. AI starts dreaming new worlds, and no two moments are ever the same," he noted, referencing Borges' Library of Babel as an inspiration for this endless variation. The result is a work that unfolds unpredictably, making each visit unique. "Visitors who return will never see the same thing twice. The chapters evolve, the logic remains, but the form is always new."
As he notes, the work actively engages with the space. "This artwork doesn't just sit in a space—it sees the space. It knows where it starts, where it ends. It adapts," he explained. The room's unique structure, with no parallel walls, presented a particular challenge. His team developed specialized software that maps the curvature of the space in real time, allowing the work to exist in constant dialogue with its environment. "The work is 360; it doesn't have a beginning or end," Anadol noted. "It's just alive."
Sound also plays a key role in this experience. Longtime collaborator Kerim Karaoglu recorded binaural sounds from Gehry's buildings, capturing the wind moving along the museum's façade and the acoustics of the interior spaces. The result is what Anadol calls "data dramatization"—a process in which AI interprets and sonifies environmental information, adding another sensory layer to the work.
Anadol's collaboration with AI used machine learning models to interpret architectural data—textures, materials, and spatial dynamics—transforming them into evolving, immersive experiences. While AI generated content, Anadol emphasized the importance of human input in shaping the emotional and visual aspects of the work. Trained on a vast archive of architectural data the artist fed into it, AI helped create dynamic environments, with Anadol guiding the emotional tone and aesthetic choices to ensure the art reflected a thoughtful intersection of technology and human creativity, rather than being merely algorithm-driven.
While training AI models often requires significant energy consumption, Anadol and his team made a conscious effort to power Living Architecture: Gehry with renewable energy sources. "If we are talking about nature with AI, but training AI using fossil fuels, then we are already failing," he stated. Technology, he argues, must be integrated with ecological responsibility, which is why his studio's AI models prioritize sustainable computation. Through collaboration with Google Cloud and renewable energy, Anadol's work demonstrates that AI can be both innovative and environmentally conscious. This commitment to sustainability also informs his exploration of time in digital art. Unlike physical materials like stone or steel, which possess inherent time properties, Anadol's AI-driven creations exist in a cloud-based, ever-changing time domain that is never truly "unplugged" or static.
Anadol sees his work not only as an aesthetic experiment but as a meditation on the future—one that asks where humanity is headed in an age where technology mediates nearly every aspect of existence. "I hope we can see beyond the shiny pixels," he reflected. In his view, AI holds the potential to unlock new realms of imagination, pushing the boundaries of human creativity. For Anadol, AI isn't just about automation or efficiency—it's about exploring the intersections of human emotion, spirituality, and cognition, blending them with the machine's creative potential.
At a time when AI is often framed as a tool for surveillance, efficiency, or control, Anadol offers a different perspective—one where technology transcends its functional role and becomes a collaborator in artistic exploration. He reframes data as something poetic rather than purely analytical, transforming information streams into fluid, hypnotic compositions that invite reflection rather than mere consumption. Yet, as AI's role in creative fields grows, so do concerns about authorship, originality, and artistic autonomy. The balance between human intuition and machine-driven processes remains delicate, raising the question of whether the future of art will be shaped by collaboration, control, or something else entirely.
On the other hand, as AI evolves, it's crucial to consider who controls the technology and how it is deployed. Anadol's work prompts important questions about the democratization of artistic creation in an era where powerful corporations and governments often monopolize technology, deepening inequalities and reinforcing surveillance, exploitation, and control. AI could offer new possibilities for collective expression, but only if we confront the power dynamics shaping its development and use.
The exhibition Living Architecture: Gehry will be on view at Guggenheim Museum Bilbao until October 19th, 2025.
