The Iridium Layer of the AI Era
FM (Friday Morning) Reflection #20
You know how scientists have theorized and developed evidence to suggest that the dinosaurs were wiped out by a asteroid or meteorite? Key pieces of evidence include a layer of iridium found in the soil that dates back to that time, and the 93-mile wide Chicxulub crater discovered in Mexico's Yucatán Peninsula.
Societies and cultures are often remembered by the creations of their time, and I wonder if we will see the same kind of trace that tells us when AI came onto the scene, exemplified by the the styling of graphics, images, and text that started to proliferate in our digital ecosystem in 2022.
This period we’re in right now has some clear markers, reflected not only in our conversations, news articles, and social media covering the public onset of AI, but also in the art and visual artifacts we leave behind. I’m thinking of the graphic illustrations on articles, publications, LinkedIn and social posts, and in the digital publishing space. I see many that have been created with AI image generators like Firefly, Midjourney, and DALL-E.
You know the illustrations I’m talking about – lots of swirls, kind of random shapes and characters, and text that looks like something but when you actually read it, it’s gibberish. And figures depicted with an atypical number of digits on their fingers, and with missing or extra appendages.
Here’s an example: I asked DALL-E to create “an image that illustrates the topics, content, theme, and concepts in my weekly Friday Morning Reflections” and you can see the result below.
You can tell immediately that the illustration above was AI-generated. The wumpy and melted shapes, something that looks like an onion (or decaying apple?), and lots of sparkling lines and random shapes in the background. At first glance it’s interesting, but then I start to question:
What does it all mean?
Can I explain the design choices?
Does it represent my brand?
I think these image generators are valuable for many things and I’ve used them too. I’m not here to critique their various strengths and weaknesses, and am very aware of the concerns that they will eliminate creative jobs and work. At least in my case, no jobs have been eliminated because, as an independent writer, I don’t have a staff nor the budget to hire a designer in the first place. So they’ve been additive for my purposes.
As a creative, though, I’m thinking about broader implications and what messages are being sent by the prevalent use of AI imagery. When we see changes in art, music, and architecture, they are reflections of what was going on in society at that time. Think of the evolution of music, reflecting changes that were going on in broader culture. Jazz, rockabilly, 80s pop, 90s grunge – they’re all markers of things that were happening in popular culture at that time.
So I wonder what the rise and prevalence of these AI-generated illustrations says about where we are today. At first glance, they can be interesting, inspiring, and descriptive, but then quite often, when you get to the details, you realize there are a lot of nonsensical graphic elements and they lack a coherent concept. Yet we use them. This suggests a general appreciation for the arts, design, and the visual medium as an important channel to communicate ideas. Because otherwise, why would we create these images?
I do think the sophistication and coherence of AI-generated images will improve as the technology evolves and people involved in creating these engines figure out how to build models that can adequately represent abstract concepts. That’s potentially part of the challenge with these engines. When you prompt, “Create an image that represents the dissatisfaction of consumers in a world where negative brand experience is threatening to break the relationship between consumers and the brands they love,” there’s a lot of abstraction there, and a lot of embedded concepts that are hard to illustrate.
Success without effort is not something that we can realistically expect. After all, we know a base LLM, out of the box, will struggle to successfully “Build me a sales presentation to help me win the advertising business of a local pizza franchise.” I’ll get an output but it will not have the brand insights, knowledge of the prospect, particular market dynamics, competitive intelligence, and other details I'd need to develop a compelling pitch.
There’s a process that I go through when coming up with concepts. It’s a process that, in my mind's eye, is shaped like an hourglass. You start at the top with lots of ideas, brainstorming concepts. Then you bring those to a point of singularity where you have the core idea. You can state it simply and clearly — a singularity where all the ideas come together in a single point of focus. And then you deliver the concept across all the different paths of execution – you take it to different places, develop a variety of strategies and tactics, all based on that core idea.
The starting point is to clearly articulate what you’re actually trying to explain or communicate.
In these Reflections, I often offer guidance and actions that you can take, or that we can do together, to help make things a little bit better.
When working with image generation engines and developing prompts, I suggest simplifying your concept to a couple of key ideas and start with those. Really focus on the idea, feeling, or takeaway that’s important. A big part of design, photography, and composition is to reduce the concept to its essentials. And that’s good for every type of communication.
Circling back to the idea of an “iridium layer” for the AI age — the imperfect images generated by AI that we often see now, with their swirling forms and surreal distortions, reflect both the infancy of these tools and our willingness to embrace experimentation.
In time, maybe we'll interpret this period parallel to other historical periods of artistic experimentation, such as the abstract movement in painting or the rise of atonal music, where creators pushed boundaries and their works reflected broader societal shifts. Perhaps they will tell a story about where we are: at a crossroads of innovation, curiosity, and imperfection.
Our responsibility today is to engage thoughtfully with the AI tools we use, ensuring that the artifacts we create are not solely driven by algorithms but shaped by our human insight and reflecting our values. And whether we work with designers, generate images ourselves, or collaborate with AI, the goal remains the same: to create artifacts that resonate with the core ideas we wish to communicate. In doing so, we contribute not only to the evolution of these tools but to the enduring narrative of our time.
Photo Bonus
This photo was definitely not generated by AI. I captured this moment on the plaza at Lincoln Center on Wednesday, freshly glossed up by the day of rain and wind. There were people walking all around but they “disappear” in time exposures like this. Captured with my Polaroid SX-70 (MiNT SLR670-X) on Green 600 Film.