Where will new computational processes take architecture and engineering in the future? We brought together Professor Thomas Schroepfer (Director of SUTD’s Advanced Architecture Laboratory and Co-Director of the SUTD-JTC I³ Centre), Dr Hossein Rezai (Director of engineering firm Web Structures) and Tan Szue Hann (Head of Sustainability at Surbana Jurong) to discuss advanced digital design processes and their take-up in Singapore.

Thomas Schroepfer (TS) Digital processes are used in multiple ways in architecture and engineering. Some of them are more innovative than others. I’d like to quote Achim Menges, a colleague in Stuttgart, who is a leading researcher in computational design and wrote an essay for one of my books. He makes a very good distinction between the various strategies that exist in computational design. He says essentially we have three main strategies. There are post-design control strategies (that is, building systems that make a building perform better); post-design optimisation strategies; and pre-design information strategies.

He’s most interested in pre-design strategies – analysing your context and bringing this information into the design process, in other words, the building form. That is radically different from the traditional approach to architectural and engineering design, where the architect is often concerned with form making and representation. That’s a very big shift in thinking and has implications on how we teach design nowadays.

Our students here at SUTD learn about the digital world in year one and subsequently incorporate pre-design strategies in their work. My research lab has developed a new computational tool that is used for design optimisation called Opossum. I think these are good examples of new crossovers – of architects developing software and making their own tools. Opossum has been downloaded by other designers more the 1,000 times over the last couple of weeks.

Tan Szue Hann (TSH) A lot of what we do these days in architectural and engineering design is guided by the performance requirements for buildings. We start running computer simulations before the building is designed in order to understand, first and foremost, how it can perform efficiently with regard to structure, mechanics, and climate response. This should be fundamental, and together with all other design considerations, influence the form and nature of the building.

Hossein Rezai (HR) I’d like to take one or two steps back – to examine our relationship with tools. Over thousands of years, since we first sharpened a piece of stone to cut something with, we’ve had a relationship with our tools. As a result, we’ve gained the upper hand over other species.

Our relationship with tools has till now been one directional and instructive. We’ve told the tool what to do. That is changing with computers. There are now tools that are being generative. This has already opened up a lot of opportunities. Now we’ve gone beyond that to the point where we are teaching intuition to our tools. We can interact with a machine like we do with our colleagues.

TSH The machine is probably not as prejudiced as human beings would be toward certain challenges.

HR Absolutely. The logic of the computer, which we have taught, is based on genetics. We call it the genetic algorithm. It is based on the principles of evolution – survival of the fittest – except that we determine what fitness in a building is. It could be aesthetic, cost, context, or whatever we determine to be the genes.

TS Many designers resist for example the use of optimisation tools because they feel they force them into quantifying problems. They prefer to keep things more undefined, particularly in the early phases of the design process. But the computer can help to identify good solutions that fit all criteria that are defined upfront, and the designer can then choose from them. However, I believe at least for the moment, the designer is still needed as the editor of these possibilities that the computer helps us to identify. I think there is an interesting middle ground to explore.

HR Some of the programmes we have are no longer tools; they’re actually working with us. And we are on the cusp of doing even more collaborative work. Looking ahead, we should change the paradigm so that our relationship with the programmes available to us will change. If we see them as collaborative then we will use them differently.

TS We’re going to see more AI and other technologies applied to work processes in the near future. What does this mean legally? What does it mean for the design profession? And the other consultants who work with us? These are important questions and concerns that are not limited to the design and building industry.

TSH Yes, and how do you programme the fact that architecture is fundamentally a cultural artefact into a computer programme or a machine algorithm? 

HR Right now a machine can’t do it so we can just tell it, ‘Yeah, you’re inferior.’ Very soon it will be able to. Driving here I saw a lot of buildings that are very culturally inadequate and historically non-existent. They’re just buildings – done by people and not by machines. We need a bit of humility.

Questions about legal frameworks are defunct nowadays. If I put something on the net, anybody can, within reason, see it. Strait-jacketing me and the internet into the old-fashioned legal frameworks is not working. Let’s try to change the legal framework to suit us.

Read the complete conversation in Cubes 87! On shelves now!