The Mesmerizing Wonder of AI Image Generation
One of my favorite activities between ages two and six was to watch people draw, and I was known to constantly pester the adults around me to draw for me. Watching a pen or pencil bring some idea in my head to life on a piece of paper mesmerized me.
This sensation returned to me a recently when I joined a Discord server operated by Midjourney, a San Francisco company using artificial intelligence to generate images based on written prompts. Using this service scratches a primal itch in my imagination. A couple things have happened to me in recent days that I think reflect its impact on my psyche.
First, I’ve had dreams about using Midjourney. My subconscious often responds to the technology I use in dreams. Frequently, I’m pulling out my iPhone in dreams to take a picture of something. I’ve had dozens of these dreams. A few weeks ago in a dream, I took a selfie with a bloated, maniacal bat that I discovered in a dilapidated house. Often times when I wake up from an iPhone dream, I’ll find myself in a liminal state in which I feel excited to check my camera roll to see how my dream pictures turned out.
I don’t remember much about my dream about Midjourney beyond an overall sense that my subconscious really liked it. Midjourney feels like catnip down there in the places where strange images coalesce. I’m reminded of the 1991 Wim Wenders movie Until the End of the World, which features a device that records dreams so that dreamers can view them when they’re awake. This causes various characters to go off the deep end and become addicted to their devices; it’s easy to see parallels with an image-generating AIs on a smart phone.
The day after my Midjourney dream I was in yoga class, concentrating on my sixth chakra, commonly known as the third eye. Ellie, my teacher, asked us to focus on a space inside our skulls a few inches from the bridge of the nose. Seated on my mat, eyes closed, I got incredibly high for just a few seconds. My overall impression was that something is going on in there, some new process I’m not entirely privy to that I had set in motion.
Here’s a theory about what’s happening. I’m using my conscious mind to interact with a tool that excels at combining and depicting concepts that are not usually combined. Creativity relies on this sort of neurological functionality, the capacity to connect neurons that might not otherwise have reason to converse. Maybe my subsconscious recognizes a kindred spirit in Midjourney and is attracted to it. It feeds me prompts that I can then pass on to the AI, which provides its own interpretations of those prompts using neural networks. While the logical, methodical parts of my mind can engage a computer in a game of chess (and be easily beaten), this level of AI provides a pathway to a much more mysterious part of the mind. The neurons in my brain conspire to offer prompts that I then use my fingers to tap into my phone. These prompts are ingested into, I’m assuming, the cloud, where they interface with another branching, iterative, algorithmic image-generation factory that turns language into shapes and colors, delivering content that I can ingest back into my brain with my eyeballs.
Midjourney arrives in my life during a period when I’ve suddenly taken up visual art. This summer I started drawing and painting, mostly using oil-based paint pens and markers. My paintings very much convey the skills of a middle aged man who took a couple art classes in high school. The portrait I attempted to paint of my partner drew withering ridicule from my teenage art critic daughter; she said it looked less like Lisa than like “an alien in a Dolly Parton wig.”
Come to think of it, an alien in a Dolly Parton wig is actually a pretty good prompt for Midjourney. Here’s how it turned out.
Mind Colonization
This time a year ago I was enthusiastically using the AI GPT-3 to create weird little hybrid short stories on such topics as an underachieving cult leader and a race of sentient beings that evolved from IoT monitors. After a few months, I grew bored with natural language processing and returned to the novel I’ve been writing all by myself for the past few years. I suspect I’ll go through something similar with Midjourney. Both platforms seem to involve an interface with part of my mind that until now I never expected to be colonized by technology, the part responsible for creative expression. As beguiling as these platforms can be, in the end I’m hard wired to find pleasure in making things by myself, from scratch.
I don’t feel threatened by GPT-3 or Midjourney and find it remarkable and beautiful that machines have begun to evolve the capacity to generate ideas through language and images. Artificial intelligence reminds me of certain animals that see themselves reflected in mirrors and appear tricked into believing they’re looking at another animal. As a species, we’ve been expressing our inner selves so long through patterns made with pigment and squiggly lines that it’s easy to not mistake these modes of expression as thoughts themselves rather than external technologies that convey them. This is an important distinction. We’re an animal that’s obsessed with tricking itself and we love falling for our own tricks— consider that some of the most revered figures in our culture are people who get filmed pretending to be other people.
When I enthusiastically approached my teenage kids with samples of my AI-generated art, their reaction was a big shrug. Midjourney is so three months ago, apparently. Yeah, uh, that’s all over TikTok. Perhaps one of the most astonishing things about technologies that replicate artistic expression with uncanny verisimilitude is how quickly we can get bored with it.
