What was the starting point of your project? The starting point of this project was exploring what a house made out of Sierpinski’s Triangle would be like. This project was primarily about getting a feel for a more structured approach to using AI imaging in the workflow of a creative design project (specifically in the context of architecture). What were the concepts and inspirations for this project? In 2018-2019, for my MArch course, I spent several months studying fractal geometry. The course was designed around exploring fractals in the context of the relationship between emerging fabrication tools and parametric design. Although the subject is highly abstract in nature, I find it quite interesting to explore how it could affect architectural technologies. Tell us about your design process! Being new to AI imaging, the process was heavily driven by curiosity. With such a powerful tool, it is easy to quickly get lost in any number of rabbit holes. It is like working with someone who is the very best at what they do, but they do not speak the same language as you. There is a lot of trial and error, but I found that the ‘blend’ tool, in particular, allowed me to have a surprisingly high degree of control over the output after many generations. Tell us about your experience designing with AI tools? And how did it help create the project? I have no prior experience with AI imaging tools, but I was able to draw a lot of experience from my computational design background. Like Rhino’s Grasshopper, it feels like a tool that simply allows you to explore ideas and iterate with various parameters very quickly. What other software did you use to curate your designs? Although I am very confident with my Photoshop skills, I opted not to edit the output images manually very much. I think that forcing myself to learn how to be more precise with the tool is more interesting for now. What is the key feature of this project that you want to talk about more? It's a super hard balance trying to create geometry that is interesting and true to the original geometry, while also creating architecture that would be nice to experience. The more computationally interesting ones tend to be cooler but not welcoming. I also find it tough to not get too obsessive about it, as so many of the results can be taken in very different and interesting directions. Conclusion: AI imaging has really exploded over the last few months, and I feel like I’m a bit late to the party, but particularly recently, people have been getting a lot more controlled and quality output out of MJ. There's a strong argument to be made that Midjourney architecture is just eye-candy; it's purely form over function. On one hand, it enables anyone to discover, combine, and iterate on high beautiful imagery at unprecedented rates. But it doesn't really do anything in the way of real-world problem-solving, which I believe should always be a key aspect of architectural design. I believe it is comparable to any technology that has come before it: ultimately it is just a tool. But it is also a ridiculously powerful tool, and I struggle to describe with words just how insane these AI tools are getting; it feels like genuine magic. MJ as a piece of image software is out of this world in its own right, but I find it particularly interesting thinking about what changes it will have on the architecture industry, whether people like it or not.