2022 was a wonderfully productive year for the design community as a whole. One watershed moment will define it, however: the introduction of AI tools DALL-E, Stable Diffusion, and Midjourney, to name the most popular and accessible of them. From June onward, a proliferation of images and discussions arose, sparking something of an infatuation that was not only contagious but revelatory. In all that hubbub, I followed the herd (or rather, I jumped on the bandwagon) and haven’t looked back since.
In a little over nine months, DALL-E and Midjourney have generated over 30,500 images (for a total of 445.35 hours of rendering time) on my behalf. Wonderful little workers, they are, never tiring or complaining. They just work and work and have gotten much better — good, actually — at rendering complex ideas realistically.
Like any self-respecting latent academic and professional architect, my initial explorations focused on architecture, pavilions, sculpture, some furniture, and the occasional landscape representation. Then, as I continued to delve deeper, that became less and less so. Ultimately, my interest and prompting efforts shifted exclusively to small-scale facades, material studies, and furniture. It is within these three areas that I have found a truly enjoyable zone for exploratory work.
Furniture is wonderfully complex and nuanced, with performative requirements tied to the body, requiring one to understand how materials affect our mind and form. In that context, furniture is an interesting area for research. It is no coincidence that Charles and Ray Eames, within their continuous feedback loop of prototyping, created countless iconic midcentury modernist pieces known as much for comfort and utility as for beauty. Their ideology was summed up perfectly by Charles when he said, “What works good is better than what looks good, because what works good lasts.”
It is with this idea as a backdrop that I have enjoyed using AI as a creative tool. Although AI has no understanding of the physical requirements of the body, it is nonetheless trained on countless images which are based on functional performance. Therefore, the primary aim of exploration is not the question of fit but rather the exploration and testing itself. AI can and does often fail both conceptually and in terms of fit, and that is perfectly fine. What one would consider a monstrosity is also okay. AI doesn’t edit away potential, and the fact that it will not push back on “crazy” is not only good, it often leads very creative outputs. In that, the AI process encourages one to blend the noncontextual and useful into hybridized forms yielding expressive results.
I cannot stress enough that these tools have enabled, through their quickness, what I once achieved only through sketching — time to think, rethink, and think again. In my busy professional and family life, I have found that tranches of 10–15 minutes are enough to be truly creative and can uncover areas to be further mined and transformed into something. One such something is a dining table, designed for members of a burgeoning middle class who aspire to own something different, special. Set to be realized — ironically, by eye and by hand — in marble by skilled craftsmen in India, this first something has an expected delivery date of July 2023.
Kevin Patrick McClellan, AIA, is an artist, designer, and principal at Marmon Mok in San Antonio.