I never imagined I could type a few words, and then four images would appear triggered by the prompt. I read about work being done with AI and visuals, particularly at Google, but this was different. It was easy to use, fast, mysterious, and surprising—nothing like the AI I had read about.
Midjourney is a text-to-image AI program that I have used since September 2023. I wanted to understand how the program generated images, but it was challenging to find out at the time: this was a new kind of AI.
When you type a prompt into Midjourney, it scours the Internet to find the Alt Text descriptions that match the prompt. Alt Text was developed for blind people who can listen to the description. Some software packages will automatically generate Alt Text, such as Microsoft Word. I generated the Alt Text for the image in Microsoft Word:
The Alt Text generated for this text in Microsoft Word is "A group of goats in the clouds."
On this website, I document the AI series I have pursued since using MidJourney. I define a series as a group of images that include a specific prompt. For example, all the works in the series Dream included the word "dream" in the prompt, and the works in the series Frida included the word "Frida Kahlo."
I found the variety and lack of any awareness of the AI-generated images to be particularly fascinating, amusing, truly outrageous, and often thought-provoking. Why did the AI choose these images? What does this tell us about the model and those who are fine-tuning the model? Midjourney v 4 had a difficult time with fingers and eyes. Those problems were resolved in later models, and now, there is rarely the wrong number of fingers, and the eyes look more realistic. When I used the prompt "Dream" in Midjourney version 4, most of the images of the women were flawless white women, and for whatever reason, there were fish in about a quarter of the pictures- I would call them hallucinations. Few women are flawless, and fish are not a common element in most dreams.
Brian Eno has been an inspiration for decades. He was part of the New Wave band Roxy Music, composed Ambient music, and produced Talking Heads' Remain in Light. Eno believes that variation and randomness are fundamental to creativity. In his view, creativity thrives on exploring variations and experimenting with slight differences. Digital art encourages variation; it is easy to change. The ability to create variations, save them, make changes, and keep making variations is a feature of Digital art.
My experiments with AI are experiments with variations. I use the prompt "Dream," I run the prompt several times. The images vary with minor and major differences. I focus my attention on deciding which images I want to expand upon. Midjourney creates four variations in about twenty seconds. From the grid of the four images, I can choose to generate variations of the original four grids or just move on to the next set.
I have always considered my computer a collaborator. I see my digital art tools not as a tool to replace human creativity but as a partner that can augment and amplify the creative process. Technology opens up new possibilities for expression, enabling creative exploration and experimentation in ways that weren't possible before. I am fascinated by the potential for technology to augment creativity rather than replace it.
I don't like the word AI. I would instead think of AI as Machine Creativity. Human creativity arises from our experiences accompanied by emotion. Translating that emotion, or passion, into art is the artist's task. The traits of human creativity include playful, curious, courageous, hardworking, obsessed and can't let go, the dreamer and doer, intuitive and following intuition, able to hold the tension of opposites and wait for a solution, large ego to believe they can do it, knows how to choose a problem, knows when to throw away and let go of ego. In other words, creativity is complex and requires many human skills.
Human creativity relies on sentience consciousness. Machine creativity, on the other hand, generates outputs using algorithms, training data, and existing patterns. Human creativity and machine creativity are very different, though the outcome of both is creative.
