Last year I finally replaced my old hand-me-down laptop with a brand-new, modest but still exciting, Lenovo Yoga. I bought this mostly for work, but also to have as my everyday machine. I expected to do a lot with it. One thing I hadn’t really foreseen but maybe should have, was how I would start working with generative AI.
“Learn AI!”
Everyone hears, every day it seems, that we need to learn to incorporate AI into our everyday lives and work. That those with expertise in AI have better prospects for future success. I personally had been more than reluctant to dive into AI. I’d been skeptical of its usefulness or its value, and I’d been doubtful of our real need for it. In addition, I felt it would only be a negative, especially for those of us who create for our livelihoods. Hearing that AI could write articles, blog posts, even books, filled me with a certain amount of dread. It was also irritating to me. This seemed largely out of our control; it would happen whether we liked it or not. My concern of course extended to other creatives in the visual arts or design communities, as well as to society as a whole.
Creating my own images
So when I got started with my new laptop, the last thing on my mind was engaging with its AI features. Well, that changed. I was working on one of my blogs, struggling with ideas for illustrating a few articles. For as long as I have been doing this – blogging (which was originally called online journaling, when I first got started at it) – I have created the vast majority of my own images. Occasionally, I will include photos, maps, or other images that are more meaningful for a particular article than anything I would design myself. But, in most cases, I have done my own art and graphics for my sites. Even when I was a columnist for a print magazine, I designed the graphics for my column and provided the images each issue.
Jumping in with Copilot
I decided to play around a little with Copilot, which was included in my new laptop. When I write an article I often have a fairly solid idea of how I’d like to illustrate it. My imagination typically comes up with several ideas, and from there, I kind of just start working on it. If it requires a new photo, I consider where I can go – if I need to be in a certain kind of place or environment. Or I consider what items I have on hand that might serve as effective “props.” As an aside, I like that this somewhat follows in my father’s footsteps to a degree, since for years in the 1960s and 70s he was a display artist for large department stores. Putting together props or items for an illustrative photograph has its definite similarities to designing and creating a storefront display.
Writing my first prompts
With Copilot I thought of these articles and what needed to be illustrated, what kind of illustration could add to the writing, could reflect the meaning of the writing; something that would perhaps not only draw people in or grab their attention, but also anchor their attention, so to speak. An illustration helps to get across and cement the meaning of the writing.
I was new to this, so I just wrote rather detailed prompts of exactly what I was looking for. And, I was quite surprised. The resulting images obviously weren’t what I would have drawn by hand (besides, my drawing skills do not go that far for illustration purposes), and if I’d tried to put together photos, it would have been nearly impossible. Since my AI-generated images included people, photos would have required far more than I have the capability of doing – setting up a location or studio, hiring models, supplying costumes, etc. I was, therefore, somewhat pleased with the AI-generated results, and they worked well with the articles.
I was also quite surprised. Not just with the results themselves, but with my satisfaction with them. I do tend to be fairly open-minded, which for me, includes being open to change, open to considering other opinions or viewpoints, open to trying new things – including those things I have previously dismissed. But I hadn’t expected to have a positive feeling from having played around with generative AI. Paradigm shift, for sure.
Illustrating memories
Deciding to play around a bit with Copilot, I thought, what if I could generate some images that illustrated moments of my childhood – childhood memories? These of course would not be actual memories (haha), but what would they look like? Would they provide any sense of genuine nostalgia for me? Would they effectively illustrate memories I had that I might share with others? I have a strong imagination and tend to visualize things fairly easily, but that’s not true for everyone. We each are different kinds of “thinkers,” so I don’t see it as a shortcoming if someone has a harder time picturing something.
A day at the beach
I thought of a few memories from my childhood. These were not specific events or occasions, rather, these were simple, somewhat generic moments in time. I first chose the kind of moment I would have had at my favorite beach growing up. My family went there often. The beach was about two hours from where we lived. We’d often go out for a long afternoon, enjoying the quiet, the solitude, the beauty, and the peace.
In my prompt, I described a little girl (representing me, obviously), how she looked, how she was dressed, her emotions at the time, and what she was doing. In this memory, I was holding a small starfish I’d found in the sand. I was quite happy, as well as fascinated in that childlike way, by the little creature. In minutes, Copilot offered an image. I had to laugh, because it was mostly quite accurate (look closely at the picture to see where it really failed). Again, it was not me in the picture. It was not the actual memory, and it was not an actual picture from that time (all obvious), but it got across my experience of such a moment. If I showed someone else I wouldn’t say, “oh, here I am at the beach when I found a starfish in the sand!” but I could say, “this is a lot like what I used to do at the beach when I was a kid,” knowing that the beach actually looked a lot like the one I had gone to, because of the details in my prompt.
On the railroad tracks
I did another, fast-forwarding some years to myself as a teen, in a much different place. This was my home, my “backyard,” and I walked alone at dusk, lost in thought, and with a degree of sadness. Again AI took my description and very closely showed me what could have been a snapshot of me at a certain age, in that familiar place that had been home. If I wanted to share a story about spending hours at night walking along the train tracks behind my apartment, I could use this (or a similarly AI-generated image) to illustrate my story. Despite the issue of the buildings being a little closer to the tracks than they really are, this image does a very good job at showing the kind of place I lived through much of my childhood.
Is it creativity?
I wouldn’t want AI to replace any true artistic endeavor. I don’t think of these images as having been my “art;” they were made to my orders, so to speak. These were not drawn or painted by me, by my hand. However, one thing that really surprised me and stood out to me, was how creative it had actually felt. There is creativity in vision, in envisioning itself. When I stand at an easel with a stick of charcoal, I don’t envision anything ahead of time. If there is a model, or an item I am drawing from still life, I suppose that is the vision – the vision is in how I am seeing what is in front of me. I don’t see a finished piece of art when I’m drawing. I let it come through me. In working with AI, maybe sitting with the chat box and creating the prompt is like standing at the easel. I’ve formed a vision of what I’d like to see, written a detailed prompt, and AI has matched it as well as it has been programmed to.
Obviously, these AI-generated images lack the soul of real art. They lack a certain depth or complexity. Many AI-generated images have a certain recognizable kind of lighting; they are a bit unnuanced. They will not evoke the way a great painting might. But they might still communicate a feeling or a mood, which in some cases is enough. For the purposes of illustrating certain articles, I have found them useful and helpful. For such purposes, I am not going for fine art.
A quick glance
I think we know that unfortunately, the average attention span has shortened greatly (are you still reading? wow, thank you!). If the average reader is just skimming an article or post, looking for a TLDR, how much attention are they paying to the accompanying artwork then? With this in mind, I have accepted the idea of generative AI for such needs. At least for now…
Our choices when working with generative AI
AI is still super new to me. I have still only played with it a little bit, and I do not use it for writing, even to clean up or edit. I won’t even use it in replying to texts or emails. And I have only ever used it a few times for illustrating articles (and always note when I have). Each of us, even if we accept AI into our everyday lives, will have our own areas where we will refuse to engage with it. But I am very curious about where we are headed with AI, and its place in our creative lives.
Thanks for reading. And as always, feel free to reach out with any questions or comments, or just to say hi.
~ Nellie
Images: main image created with my prompts for a wedding in the woods, inspired by my Polish heritage; others inspired by personal memories. Created using Microsoft Copilot and Photoshop.
