Artificial Intelligence and the Authorship of the Architect
Our technological evolution to the post-digital age (Berry and Dieter, 2015) has resulted in an unprecedent challenge to the authorship of the architect within the design process (Ajhua and Chopson, 2020). Improvements in computer processing power have provided alternate means of decodifying complex design problems through generative algorithms which mimic the evolutionary processes of nature (Cubukcuoglu et al, 2019). Fitness functions deliver idealised design outcomes, within the context of complex parameters and variables. This process of generative design mimics the architect’s own iterative design process by searching for an idealised outcome. Here, I intend to explore the impact that these automated processes of design have on authorship in practice, explored through the framework of case studies that quantify the range of interaction between the algorithm and architect.
The establishment of the Royal Institute of British Architects (RIBA) Code of Conduct (Rule 3.1 2019), in combination with the Copyright, Designs and Patents Act (1988) have provided UK architects with a ‘safety blanket’ of regulatory and legal tools to challenge the unlawful reproduction of part, or all of their designs. The battle for innovation in design to win work has historically taken place within the architectural community, observing the profession’s framework of fair competitive standards. This confidence in authorship, upheld in law and governed by regulation, has arguably led to complacency in the self-perception of the architect as ‘master builder’ within the design team; a role devised as early as the codified drawings of Alberti (Watkins, 1957).
The challenge for the future of architectural design now comes from outside the profession; mathematicians, engineers and computer scientists are increasingly automating complex design processes by generating idealised solutions against fitness criteria, integrating structural calculations and testing fabrication limitations in parallel. The total automation of architectural design may be a real and present threat to the profession; already intimated in part through the creation of accessible software (Liwicki et al, 2017).
There are few sources of theoretical text, articles or projects that refer directly to authorship in architecture. Saint’s book the ‘Image of the Architect’ (1983) presents an historic overview of the identity of the architect, challenging stereotypes in film, writing and text as the figure of Hero, Gentleman and Entrepreneur, amongst others. Chance’s ‘Fame and Architecture’ (2002) similarly focuses upon identity but explores more directly how envy leads to fervent protection over intellectual property and innovation.
The book “Architecture and Authorship” (Anstey et al, 2007) is perhaps the main work on the subject, comprising sixteen essays that analyses contemporary authorship through individual and diverse case studies. As an indication of the rapid speed of algorithmic design adoption since 2007, there is just one mention of authorship in the context of generative or ‘evolutionary’ algorithms, p.137 “..in some respects, evolutionary simulations can be said to replace design, since artists can use this software to breed new forms rather than specifically design them..”. Discussion in the chapter ‘Cedric Price as Anti-Architect’ details his 1964 pseudo-algorithm design for the Fun Palace, and a later chapter “Systems aesthetics, or how Cambridge solved architecture” refers in part to how Alexander battled Eisenmann to champion computer over heuristic design.
Carpo (2011) in his book “The Alphabet and the Algorithm” talks more explicitly about the potential for generative design methods in architecture. The work goes into some detail on the paradigm change in architectural authorship between Alberti and Brunelleschi in the construction of the Firenze Duomo, an important case study in defining the scope of work of the architect. In his final chapter ‘the Fall’, Carpo concludes by imagining a future where architects must adapt to survive and become “the programmers and masterminds of nonstandard series, they will preside over the full extent of general, genetic or parametric visual environments at all scales.”
The concise “Authorship in Architecture in the Postdigital Age” (Ortega, 2017) published in the Total Designer series of books, provides a summary of the historical and theoretical developments in the notion of architectural authorship from medieval to contemporary practice, speculating on a future that implores architects to take control of the algorithm. Ortega poetically concludes his book by calling upon architects to reconsider their role as autocratic authors in the face of algorithmic revolution in his ‘Manifesto on the Expanded Designer’.
Despite the voices of Carpo and Ortega rallying architects to take control in the face of automated processes of design, there is currently no quantative or qualitative research completed on how these automated processes impact the role of the architect and their designs now and in future. There are no illustrative or methodical case studies that descriptively test or illuminate the proposed area of study. A key observation from the literature review is that many papers written in the area are completed by computer scientists and software programmers, who are automating the processes of the architect from a background outside the subject.
In the context of this gap in knowledge, this work will first attempt to quantify the notion of authorship of the architect in practice. I then intend to formulate a new methodology to visualise how authorship changes through case studies that test the limits of algorithm / AI / architect design interaction. This will be achieved through a shared diagrammatic framework (Hensel et al, 2004) that visually illustrates the differing design process in relation to the role of the architect.
The work will make explicit the opportunities that this paradigm shift represents in the design process, to inform the education and regulation of the profession in future.