Rethinking Automation

In the mid-19th century, painter Paul Delaroche declared, "From today painting is dead!" after seeing some of the world’s first printed photographs. He felt that these effortless illustrations of reality would destroy the need for craft in image representation. During the same period, at an early meeting of the Photographic Society of London, another group of people – far more sceptical this time - described photography as, "Too literal to compete with works of art."

If Delaroche’s idea had taken hold, we would have witnessed the failed careers of Picasso, Kandinsky, Matisse and other celebrated painters from the 20th Century. Artists would discover they had nothing to say in a photographed world. And if the Photographic Society of London’s scenario were true, Instagram would be a tool of scientists, rather than the creative hub that we know today. Two centuries later, we know both of these scenarios are far from true. Photography hasn’t replaced painting, and it certainly isn’t considered to be an irrelevant form of art.

In architectural design, we’re seeing similar debates emerge in social media circles, but this time it’s around the involvement of computers and automation in design processes. By ‘automation’, I mean algorithms that have been made to support and enhance the core design process.

We’re encountering people who are vehemently against any notion of computers being involved in design and who judge the results as soulless, impractical or irrelevant. Some communities are even going as far as suggesting that architecture design is in danger of becoming useless.

Architects and the adoption of technology and automation

Let's begin by saying that, as architects, we’ve always been cautious about adopting automation. There are a few reasons for this.

On one hand, when it comes to technological innovation, there is a tendency to take a reductive perspective; visionaries focus on a single point rather than considering the challenge in its entirety.  

For example, in the 19th Century, people imagined that technology would be a sort of cure-all, remedy for everything. With education, innovators predicted that technology would allow people to download information straight into brains through wires. And with architecture, they suggested that architects might be able to construct a house via the touch of a button. In theory, these ideas are nice but in reality they don’t work - and this is because they haven't considered these processes holistically. As a result, and perhaps not surprisingly, this perspective has led to a difficult relationship between architects and technology.

On the other hand, as architects, our failure to evolve and explore different ways of working might be limiting, stopping us from finding new and interesting approaches to architectural design. Arch Peter Eisenman even went as far as describing technology as a “cruel tool” that inhibits creatives.

But perhaps we’ve overlooked the creative potential of automation. Perhaps we’re approaching automation in design in the wrong way, similar to how 20th century artists saw photography. Instead of resisting it, why not embrace automation in design? We could see it as a way to create better and more interesting design procedures, a mechanism to play and to be creative, and to open different paths for designers.

Parallels in art and music

To get an idea of how automation can influence architectural designers, we can analyse the ways in which automation has been introduced to art and music. The rise of generative art shows us that it’s possible to delegate certain decisions to automatic processes without losing any creative value as artists.

In generative art, artist use software like processing or p5.js to create rules and procedures that are analysed by autonomous systems that create unique pieces of art. This way of thinking about art - as a human-computer collaboration - brings another unique layer to art that we can explore, one that doesn’t undermine the artist’s own creativity.

And music is particularly interesting because music composition shares many similarities with architectural design - both disciplines involve using objective and subjective parameters.

One of the first authors to benefit from automation in his compositions was American composer David Cope. Frustrated from a composer’s block for a Bach’s chorale he was paid to produce, he started analysing typical patterns that were present in Bach’s compositions. This revelation helped Cope to compose an entire musical system inspired by Bach, instead of a single song. With the musical system defined, he could automate the composition process and navigate through an infinite sea of chorales play with and listen to.

Introducing automation into architecture

We’ve seen some similar examples in architectural design, where quick-thinking individuals have adopted similar methods of automation.

When designing social housing in Malagueria, architect Alvaro Siza wanted to find a way of creating personalised designs for each house. To do this, he created a series of design principles that could be replicated by architectural drafters, and which Arch Jose Pinto Duarte later used to create an automated system for the whole procedure. This method combined Alvaro Siza’s artistic style with methods of automation to achieve his goal of personalised mass housing.

Taking the idea to another level, automation can also enable us to work with communities and get feedback from them. Decoding Spaces has worked with multiple communities in Ethiopia. Due to the fast pace that these communities have grown, a system to design, explore and exhibit new design proposals was needed. The automation processes they created enabled the community to participate in the design of the master plan of their cities and allowed the designers themselves to get instant feedback from the people they were creating for.

Pushing automation in architecture

At matterlab, we've been actively seeking out combinations of design and automation, trying out different methods where automation can open new spaces for creativity.

In a project for the Autodesk Generative Design Primer, we explored using generative design processes to create entourage elements to render scenes. We wanted to use generative design not as an optimisation tool but instead as an exploration tool. We started by using these processes as a way to create interesting scenery. As architects, being able to play with the results, and even rebel against the initial suggestions made by generative design helped us to find more interesting results. It was vital forgetting the creative solutions we needed.  

unitize is another example. Here, we wanted to find a way of creating a communication channel between the designer and an automation system. What we created offers users the ability to sculpt a building while automating the process of placing units in the imported shape. It also offers an instant feedback loop for the designer, where they can design a system and receive real-time analytics from it.  

‍