Musings on the Future of Design & AI
I have been stewing on a few thoughts about my use of various AI tools in my process and indirectly the wider adoption across the design industry as a whole. And while the most poignant detractors are beginning to speak more loudly and articulately concerning the value of the tools and those who adopt them. It seems we are in the midst of a continuous inflection point that will end when we are able to accept, learn, and grow with the tools, because they are not going away. Here are my three main avenues of thought that I would like to share. I offer them as aspects of the conversation that I am either concerned about or interested in.
Where credit is due
Probably the biggest elephant in the room, for me, is the notion that not all users of AI tools have good intentions. That my attempts to put those tools to work for me, my clients, and ultimately to the benefit of end-users is done with transparency and honesty is something I pride myself on. Whether I find myself inspired by biomimicry and natural structures, a particular material or finish trend, or from an AI-generated set of outputs assembled from my own ideas, a critical part of my story telling is the journey of how I got to my result.
Possible inspiration sources
It doesn’t take much imagination to conclude that pressures to shortcut the design process to get to a result faster (and cheaper) is very tempting, and to many there lacks an immediate cost in doing so. I worry about that. Not too long ago I built myself a Midjourney style tuner using the terms “H.R. Giger” and “Zaha Hadid” in the prompting. It was an experiment for myself and not for paid work, but as I did so I felt a bit icky - and not because of the creepy imagery that prompt generated. The moment I set Midjourney on that path, I realized how easily ill-intentioned users could “rip-off” others’ work with incredible ease. I decided to dig further and extrapolate what it was about those prompts that I wanted to investigate, going a bit further than asking Midjourney to reference the acclaimed work of others. I felt more content with this as a path, but in my head I knew what had inspired me.
Not exactly what I was looking for anyways.
So here’s the question for discussion. How should designers give credit where credit is due? I am of the opinion that we most definitely must report when any prompt has contains a named reference or artifact in the case of an image prompt. Beyond that, are we satisfied with telling the story of what inspired us even if not named explicitly in a prompt? What if ChatGPT is asked to describe a Zaha Hadid building and those results are used instead? What then?
Human Element
Looking at the present state of generative AI tools and the near to mid term future of their use in the designer’s tool kit, I tend to think even further out to the future impact of our collective anthropology. Beyond the historical backdrop of wars, cultural upheaval, news cycles and content strategies motivated by clicks and advertising dollars, etc., what will we make of our current cultural contribution as a society and what we value as evidenced by our artifacts? I am concerned that handing the keys of people-centered design to AI would lack a large degree of wisdom and dilute what we - the creatives solving for nascent human needs with empathy and understanding - are driven to our creative fields to do: Design for meaning.
Even the machines can't process this future.
I think about Edward Tufte’s book Beautiful Evidence and try to work out what evidence might be inscribed in the design of an object or experience that a human had a hand in the process that drove its creation. If mastering design is akin to defining it’s form, materials, expression, usability, color, attitude, relevance, manufacturability, and sustainability, surely meaning is just as important. So here is the next question: how do we ensure that we retain a “designed for humans, by humans” both in process and in the produced product? Do we expect that AI will begin to know us better than we know ourselves and let it do all of that “heavy lifting”? What do we want the future to know about us and what we valued when looking at what we produced?
Creativity is King
While not as ethically concerning as the above, the value of creativity is no less important as we adapt to the reality of AI-powered tools near ubiquity (and continuous improvement). I had a conversation a few month’s back with a colleague about the use of generative AI in the design process and how the tools really struggle with new-to-world innovations. The example I use is the imagining the iPhone in a world where such a thing doesn’t exist - or any such product improvement that substantially differs from the accepted and familiar configuration, form or manner of operation. I have touched on ways to trick Midjourney through workshopping a prompt with analogies and metaphor in a prior post. This isn’t always easily done and, at the time of this writing, is still more effectively done by hand without an AI assist.
Same Prompt for both. Who did it better?
The missing ingredient is the human creativity that a good designer possesses. That is more valuable than ever right now. There is no AI tool that can currently replace a designer’s lived experience, observation of the world and people around them, and the talent for connecting the pieces in new and interesting ways that weren’t previously explored. My old boss, Daniel Streng, instilled this value in me long ago:
Never let the shortcomings of the tools compromise the design.
So what value then are generative AI tools to such a talented visionary of a designer? My primary uses of generative AI include two main activities. First, I use the tools for various thought experiments that inform my early ideation work. For example the Hadid-Giger mashup I described earlier. With traditional design tools, there is no way to efficiently do that without feeling like you are going down a time-wasting rabbit hole. Generative AI gives me the ability to let a muse play out and decide if it is worth exploring further or not. Second, I use AI to speed up post-production of rendered imagery. Touching up renders used to take a lot of time and still does, but now I can explore a broader palette of settings, lighting and context that I didn’t have time for before. AI adds value in these contexts with out stepping on the human component.
My final thought is that perhaps it would be of value to begin developing a set AI design heuristics that would allow designers to extract value from the tools and do so responsibly. Just as usability and interaction design do. Is this a possible starting point? Please let me know your thoughts.
