Chapter 8: Post-Norm Architecture
Fabio Gramazio & Matthias Kohler
Back to Humans I
Since the completion of the Gantenbein Winery façade in Switzerland, which we designed and robotically fabricated in 2006, advances in sensing technologies have made digital fabrication more sophisticated and flexible, allowing for direct human-machine collaboration. Based on this paradigm shift, in 2019 we designed and fabricated a sequel of the Gantenbein façade for the Kitrus Winery in Thessaloniki, Greece. While the earlier process was relying on an industrial robot that placed intricately rotated bricks to create bespoke wall elements in a factory-like manner, the recently completed project re-engages local craftsmen to place local bricks in situ by construction workers equipped with Augmented Reality devices. From an industrial perspective, this sounds like an economical and technological setback. However, quite the contrary is true; this collaborative model, where human and machine start to dovetail their respective advantages in a flexible and hybrid process, is not only socially sustainable but also economically more viable than full automation. It overcomes the rigid, hardware and capital-intensive factory automation and replaces it by flexible, software based and low-cost digital fabrication technologies.
When building the façade of the Kitrus Winery the craftsmen are making use of Augmented Reality devices aiding them to manually place the bricks following a three-dimensional digital blueprint. These devices allow them to not only to precisely place the bricks but also to lay a mortar bed of varying heights. Automating such complex tasks with robots would be possible, but compared to humans the dexterity of robotic systems remains poor and mechanistic. Instead, working with a soft material like mortar is easy and intuitive for humans. The craftsman and the machines involved interact not only in a shared physical space but also, and more importantly, interact in a shared virtual space by transacting information. There is a closed feedback loop between the craftsman and the machine counterpart; the camera and the human eye concurrently scan the bricks; a screen depicts the deviances which the hand adjusts to what the machine foresees and thus creates a novel physical reality brick by brick. In the same way as Augmented Reality techniques allow for a deeper integration of humans and machines in construction, Artificial Intelligence holds the promise to do the same for design. The potential benefits are immense. However, since the architectural design space is greatly stifled by a large number of normative restrictions, in order for this human-machine collaboration to be fruitful and architecturally meaningful, the current concepts of norms and regulation need to be critically revised.
Back to Humans II
The extensive success of the Industrial Revolution is based on the appointment of, and continued investment in norms and regulations. This is not to say that they did not exist before, but norms allowed the guaranteeing of quality and common standards in a rapidly and radically changing social and economic context. Without norms, today’s globalized, digitized and fundamentally networked society would be unthinkable. The success of norms and regulations has become so fundamental that it became an instinctive societal reaction to manage complexity and organize it in simpler terms. Architecture and global urbanization are as well deeply dependent on them, despite generally not being written by practicing architects. The normative power of building codes spans from urban massing and functional zoning down to kitchen layouts. In consequence, one of the professional skills of the architect has become the mastering of building norms and regulations (and in particular, looking for creative loopholes). This pervasiveness has created an ideal context for automation– such as for example robotic task handling or generative design – to flourish. In the near future, advancements in artificial intelligence may thrive within this trend. They will potentially alleviate many of the burdens of designing and constructing and put forward many functional and logistical benefits. However, since there is more to architecture than simply being efficient and norm-fulfilling, such developments will also ask for critical and holistic agency. Not only because it is humans that inhabit architecture (for the greater part), but because for now, it is also humans that must accompany and participate in its design and construction.
Artificial intelligence has been around the last forty years but has only proven to be effective and highly impactful recently with the availability of massive computational power. Artificial intelligence is very effective in learning from large and complex datasets, including aforementioned norms and regulations. It could therefore not only help automating construction but also allow to automatically generate viable design solutions for the bulk of architectural production. This sounds very promising and attractive because it guarantees the fulfillment of quantifiable benchmarks while ensuring building norms and regulations are met. However intellectually challenging this may be to architects, these developments pose a provocative yet exciting inquiry into machines’ increasing competence to design autonomously. What happens to architecture when it is not driven anymore by architects alone but by AI software? As norms can only represent knowledge and experience that has already been put into practice – of the past, and therefore possibly outdated – it is at least questionable to automate the production of architecture solely based on them. If done blindly this would lead to endless self-replication of efficient, standardized solutions probably unfit to tackle the challenges of the future. These solutions are intellectually trapped in norms that reflect paradigms of the past and therefore they are not naturally adept at addressing contemporary design questions. This presents us with an impasse and a challenge at the same time, because the use of AI in architecture questions the cultural and historical foundations of our profession; it requires critical self-reflection. Since architecture is a human centered activity architects are not keen to automate. We are not yet used to delegating parts of our supposedly creative work to machines. Understandably, we fear a loss of autonomy and an impoverishment of the intellectual foundation of architecture. While ignoring the significance of AI could result in an even bigger threat to architecture, a proactive and creative attitude towards this technology can in turn be at the core of a positive redefinition of the human role in our profession.
One approach to strengthen the role of the architect in the ongoing trend towards automation would be to break the alliance between the superhuman capability of algorithms and the increasing density of building norms by radically deregulating the construction industry. However, falling back into a preindustrial paradigm fails to leverage the potential of the digital to structure and coordinate complex realities while abandoning the ability to help develop further the role of the architect. A radically different approach than questioning the concept of norms is to update them. Could we imagine algorithms to take the function of norms in the 21st century? What would this shift entail? Instead of norms being static, hard to oversee and sometimes overlapping or contradictory, algorithmic norms would be dynamic, transparent, local and adaptive. Imagine these norms as plug-ins to design tools, which check regulatory requirements and can provide real-time feedback to the architect; such as geolocation, topographical survey integration, height and footprint limitations, required façade and glazing treatments, etc. These norms would always be up to date and contextual because they are dynamically linked to digital twins running state-of-the-art simulations; for example on climate and energy as well as to real-world trackable variables, such as local temperature, pollution and noise maps or mobility patterns. The possibilities of contextually dynamic constraints are virtually endless. But most importantly they allow an intuitive and interactive handling of the complex and evolving requirements architecture has to meet. This leap beyond static norms frees architecture from an artificial straight jacket of standardization and homogenization. The promise of such computational norms is that they enable a dynamic, locally specific, contextual, effective and sustainable approach to complex societal challenges. This in turn requires future architects to design computational processes with the same ease, creativity and ambitions as they design the material and the social reality of architecture itself. Beyond their application, the very process of defining algorithmic norms could turn out to be a creative humanistic exercise – a driver for change in architectural culture.
Back to Humans III
Concurrently to humans that play computer games, now AI algorithms play games designed for humans. Yet to imagine that the machines’ ultimate destiny is to imitate humans unveils a lack of fantasy and propels the self-fulfilling prophecy of the machine being a competitor to the human. However, from a truly post-anthropocentric perspective, the machine has its own autonomous identity. It can learn about complexities and thus co-author architectural proposals that are beyond human capabilities. Its algorithms allow the exploration of design spaces that were formerly inaccessible. Given not only the planetary, societal and environmental challenges, but also the local complexities we face with every design project, it only makes sense to leverage on these novel capacities with a humanistic sensitivity and cultural openness. Rather than safeguarding professional architectural traditions, we shall address these new challenges with appropriate skills, and develop a novel architecture that is meaningful, impactful and resonates with our times.
Zurich, Switzerland, 31 March 2022
copyright Fabio Gramazio & Matthias Kohler 2024