Q&A with Adrian Short, Project Lead for GTA-i at Gebler Tooth
How is algorithmic software allowing for more design possibilities in architecture?
The following is a direct quote from Gehry Partner’s website;
‘The firm relies on the use of Digital Project, a sophisticated 3D computer modelling program originally created for use by the aerospace industry, to thoroughly document designs and to rationalise the bidding, fabrication, and construction processes’.
I’m not sure I would have used the word ‘relies’. But the point is well made, there are world leading practices using Algorithms and computational programs in their day to day work. As Gehry’s practice ably demonstrate in their built work. The benefits of computational Algorithms are clear to see, creating buildable forms not necessarily conceivable without the aid of such programs.
What is the best way to balance the use of software/algorithms when designing?
This really depends on what the software/algorithm is. It has a valuable role in the early stages of spatial analysis, 2D & 3D, being able to cycle through a significant number of plausible design pathways.
If it is something that generates many options - the software can be used for initial inspiration – see what works and what doesn’t. The designer can combine solutions together, build on what the computer has produced etc. Such an approach may sit under the use term ‘Generative Design’.
Other software can have a specific role – creating complex roof shapes that are constructable, for example. Allowing a designer to try out visually interesting forms in a fraction of the time it would take them to work out manually. Sitting more commonly under what is termed ‘Parametric Design’
Algorithms are simply a tool to be used, in the same way CAD is a tool for drawing. Their purpose is to help the designer make better and more informed decisions; they are not intended to replace the designer.
It is also important to note that what you get out of software / algorithms depends on what you put in. One needs a good understanding of the brief and what inputs to use in order to get a good answer.
Which elements of the design process are improved by algorithms?
Computers are great at completing repetitive tasks and can process considerably more combinations than a human would have the time to try out. By using software, designers can cut down on the monotonous, mundane tasks and focus on being creative and innovative.
How does the designer keep control of the creative process?
""Until we get to a point where algorithms replace designers (which may never happen), algorithms will only be practical if they work with humans."
Designers don’t follow a linear process, so the machine is never likely to replace the human designer. Although depending upon the design task at hand the human may be more of a curator / editor in some instances, than hands-on, sleeves rolled up designers.
Is there a possibility where the algorithms design a complete scheme? Would this be a good thing?
Algorithms can certainly generate a solution or set of solutions. Whether these are suitable for the purpose is another matter, or fully resolved.
There may be variables that aren’t designed into the algorithm, whether this is to minimise running time or because the variable is difficult to quantify, i.e. how a space makes a person feel. Is this a good space or not?
Additionally, the idea of what is considered beautiful will differ between individuals and will change over time. Humans are creative, finding new ways to innovate and new approaches that make their designs unique. This is something that computers cannot do, they need to be given rules in which to operate or shown many examples of existing solutions. An example of this might be fashion. High fashion is to most an artistic expression often devoid of practical utility, but delightful to the contemporary eye. It is hard to consider a machine being able to have the ability to create something new and emotive without the core capacity of emotion itself. The plot of many a good Sci-fi movie.. Ex Machina, et al.