Refik Anadol, Living Architecture: Gehry 

A review of our experience of the In Situ kickoff exhibition at the Guggenheim Museum Bilbao

Artificial Intelligence (AI) art has been around for longer than one might expect. Its beginnings date back to the 1960s when visual artist Harold Cohen (1928–2016) developed an AI programme that generated output through plotters and painting machines, which interpreted the commands to make line drawings on paper with automated pens, and added colour with brushes. 

Fast-forward about 60 years to find Turkish-American media artist, architect and AI art pioneer Refik Anadol exhibiting his latest work, Living Architecture: Gehry (2025), part of the In Situ exhibition series at the Guggenheim Museum Bilbao. No plotters or painting machines aided the creation of this AI art. 

Instead, 46 projectors cover the 13-meter-high walls in imagery, with a resolution of 20,000 pixels horizontally, resulting in a grand and impressionable experience. Living Architecture: Gehry unfolds across six interconnected chapters, each visualising a distinct stage from data translation into architectural imagination. Each chapter is visually unique within every cycle.

Due to 35 million images of data stock prepared for the exhibition running through October 19th, the visuals will not be repeated once or show any duplicate renders. This results in an incredibly immersive and hypnotic experience, in which the viewer constantly tries to make sense of the vast colours, shapes, and imagery the AI seems to be combining and restructuring in real time.

The sound accompanying the visual artwork is created with software developed by Refik Anadol Studio, which he calls “Data Sonification.” Anadol has added these samples — mostly recorded in and around the Guggenheim and the city of Bilbao — to the data stock for the AI to reference and create a soundtrack from.

The exhibition is undoubtedly aesthetically pleasing, awe-inducing, and borderline hypnotic, providing an otherworldly experience for audiences. It is easy to get lost in the smooth-flowing visuals and enter an alternate reality where none of the real-world issues seem to matter.

Unavoidably, at some point, the viewer has to leave the exhibition space, and with it, the alternate reality. Having to return to the complex world we all live in feels deflating, having taken little away from the artwork other than a break from reality. The immersive nature of the exhibition, along with the expansive “imagination” of the AI (as Anadol described it during the press conference), leaves little room for interpretation or personal relativity, causing the audience to be reduced to nothing but spectators.

Instead of attempting to connect to his audience through the expression of shared experiences, worries, or struggles in his work, during the press conference, Anadol discussed how he considers AI as more than a tool, since the cameras scan the geometric shape to help the software perceive and optimise itself in real time. This process sounded impressive, and still, it is just one of the ways Anadol’s technical ingenuity is demonstrated. 

That being said, the artist seems so absorbed by the technical possibilities that the connection to the human audience is neglected. The anthropomorphisation of AI raises questions about whether the people operating it are artists or technicians, romanticising their creations and losing touch with the core of human experience.

Terms like ‘dream’, ‘hallucination’, ‘perception’, and ‘consciousness’ – usually associated with the deepest layers of the human experience – are now also used in conversations about machine learning. This not only bears the moral dangers of emotional attachment and empathetic treatment of a machine but also projects attributes we as humans don’t even fully comprehend on a scientific level onto an inanimate object.

During the press conference, Anadol was confronted with a question regarding the role of technological aspects as focal points rather than transitory elements with which to communicate concepts. In response, he mentioned how the evolution of AI will impact the notions of privacy and free will, bringing about a needed restructuring of morality as we have known it so far. 

The artist also emphasised ethical data sourcing and appeared very transparent whilst explaining where he sourced the vast amount of imagery from. Before starting the project, Anadol contacted architect Frank Gehry to present his ideas and receive approval to use Gehry’s sketches and archived material from the Getty Research Institute as data stock. With the treatment of intellectual property posing one of the biggest questions in the field of AI, it felt reassuring to hear the artist’s transparency and awareness. 

Whilst divulging about the artwork’s development process, AI’s energy consumption was another problem Anadol seemed mindful of. When asked about energy consumption, Anadol told us that the entire exhibition uses as much energy as it takes to charge four iPhones in one year. This was impressive, given that numerous reports have stated AI platforms, such as Chatgpt, have large carbon footprints – using thousands of litres of water to prevent the servers from overheating.

Another element Anadol has thoughtfully considered in the development process is how the software retrieves and stores data, differentiating it from its contemporaries. Instead of storing the information, it is explained to the software metaphorically as though being read from a book, which saves a lot of energy. 

AI is now all around us, but especially in art. The purpose of the expansion is not always clear, as AI art often seems plump and shallow. Is it being used as a catalyst for positive change or solely as a means to aimlessly expand and optimise? We currently live in a world that suffers from the latter and seems to be collapsing by the day. Accordingly, artists with the exposure and funding Anadol enjoys need to use their output to drive forward and create change for a more positive future.