Future of generative design?

Counterword by Bill Allan:


Any constructive thoughts?


not sure this topic has a future


1 Like

At school I would be followed and pestered by certain persons who would put ping pong balls to their eyes and call out “ahhhh, grasshopper!”

:rofl: :thinking:

1 Like

I have seen on the internet a conference by three professors in an american university showing the project of the World Trade center Twin towers. And they laughed about the architecture they think to be a huge ordinary office building.

They laugh, but I laugh about their ignorance. It was just a very subtile architecture, mixed with discreet but revolutionary engineering new concepts.
You have two period in tall building history about technics, from the first pyramid to the Twin Towers, and from the Twin towers until now.

And when we know that the starting point of all this innovation is a design that takes into account the fear of the height of the architect…

Interesting points on both sides, and both a good read, thanks for posting. It’s worth pointing out that both these articles are focused on the generative design process in architecture, designing buildings. Designing spaces for humans to use must consider the users emotional experience in all it’s human irrationality, a hard thing for AI to grasp. I personally tend to think of architecture as sculpting in the medium of emotion, and space and material as the tools to do that sculpting, buildings as resonance chambers that are designed to focus the frequency of emotion to a specific pitch. That aspect of the design takes vision, experience and, for lack of a better term “Art”. It relies on some creative process that is unique and particular to the individual, one that nobody, often not even the artists themselves really understand much about. I guess I’m arguing somewhat closer to the Daniel Davis point that architecture is more art than science. Not all of the work of “architecture” however is the grand vision, trying to optimize HVAC, finding how many rooms can fit the basement, optimizing travel times within a hospital layout, these are design processes that are driven by quantifiable data and lend themselves well to help from generative design. Many of Bill Allan’s counterpoints highlight these aspects of architecture, where one is specifically trying to optimize a particular parameter or quality. I think in these cases there is a compelling case for generative design to be included as a tool, always subject to human editing so that the emotional component retained, so that the form can never get lost in a mindless quest for perfect function.

What is not covered here is generative design in more purely function driven applications, what we might call Generative Engineering. Designing parts and mechanical assembly by inputting the physical needs and letting the algorithm cook up an optimized shape that performs the required function with a minimum of material. This process combined with new manufacturing techniques like multi axis machining and 3D metal printing yield some amazing, sometimes counter intuitive solutions, to mechanical problems. These fluid, bone like, designs that resemble biomimicry often have a beauty all their own. This is an exciting aspect of Generative Design to me, where the subtle human psychology of architecture that an algorithm does not appear well suited for are not a part of the process and an AI can focus on doing what it good at, crunching data and returning ideas quickly.


Well, this is not my point of view.

Generative design or algorithmic design is a little open definition.

For me, algorithmic (or generative) design is :

  1. design algorithm
  2. draw simple drawings you need as point, lines, surfaces eventually
  3. plug together simple drawings + algorithm

The result is fast and extraordinary. The “generative” results or design options is not made by algorithm, but by human who modify the simple drawing, or the algorithm or the parameter. Then you get a fantastic tool to research on architecture and engineering.
This is not what I see on articles, that is a nightmare for me.

Tell me more about why you feel this way. I would like to understand.

This is where topology optimalisation comes in play. Basically, as the designer, you determine the design space and how it is gonna be used and the algorithm calculates which particles stay and which won’t be needed. This is how the architecture of Zaha Hadid’s office work.
More examples:


Again, like generative design, started in the early 80’s or so, so nothing new under the sun and actually used by nature all along:


In art, architecture, science research, politic, you have to think about the complexity.

That is to say lead to a synthesis with many parameters.
You can do it by intuition, or by dazzling after having worked a lot.

You have to work a lot.

This is where algorithmic design comes into play. It is a very powerful means and not an end.
I have the feeling that in the articles cited, we confuse means and finality.

To arrive at this synthesis, no solution works like randomness, the search for automated optimization, simplistic ideas or Cartesian thinking only.

No scientific or artistic discovery has been obtained in this way.

You have to work a lot, experiment, for the illumination to come.

1 Like

Maybe there is a misunderstanding on my part, but topology optimization is an application of algorithmic design, it is not algorithmic design.
I think that no artistic creation can be obtained with the method which is presented. I have doubts about the application presented as current at Zaha Hadid.

By overly directive methods you prevent the creative process … because it is a process, hence the interest of algorithmic design

Very cool design examples. I watched the video as well, and could only just hang on to the subject with my terrible german but enjoyed it. I think Zaha Hadid’s examples are a good combinations of the same point I was making that the generative design plays a huge potential role in the structural engineering and that can certainly allow for amazing forms, but it’s a curated process still led by human with a visionary idea.

Thanks for starting the conversation, it’s very interesting.

I think we on on the same wavelength here.

I would love to see Altair generate a SketchUp extension for optimizing topology / concept generation.

In general, I think generative design should be a tool for the designer, it can never run on its own.

Then again, I can only see so far in the future…