AI Project: The Tethys Expedition Logs

Photopolymer etching (above) & Digital images, 12.5 x 12.5 cm; Video, 5:46 min. 2024

In 2024, I participated in “The Ascenti AI Project”, which explores the question: How might professionals in the creative arts collaborate with AI programs to discover new tools and techniques for their work? The overall project encompasses a book titled “Ascenti: Humans opening to AI”, a dedicated website at ascenti-project.com, and a conference symposium.

The AI images above were generated with prompts like: “On deep dark seafloor, thousands of deep cerulean blue and purple steampunk trilobites are harvesting glass sponges, steaming sea vents in glowing bioluminescent lights in the style of a Eveline Kolijn print.”

I embraced the Ascenti opportunity to experiment further with AI and ponder on implications, surprises, failures, and what uses it can have for me as a tool in my practice.

My first goal was to see what AI would generate when trained with images of my prints. Second, to turn an AI-generated image into print. Once I entered reference images of my work and entered prompts in a stream-of-consciousness style, the results were more connected to my work and quite mesmerizing!

With that amount of eye candy, I asked myself: “Why would I need AI to create fantasy sea creatures, when I can do that?” It made more sense to ask AI to generate components I am less proficient in or can’t do. I adjusted my plan to create a video, in which AI would animate a selection of my 149 generated images, and generate sound effects, music and voice-overs.

The following AI tools were used to make my video: text-to-image generation, generative fill and generative expand, depth map movement, text-to-video generation, image blending, text generation, song and music generation, sound generation, voice narration, watermark removal and video upscaling.

I made a short video on the first steps of my process:

The script of the “Tethys Sea Expedition Logs” emerged organically, by stitching scenes together. A simple trope of an underwater expedition exploring creatures in an unknown marine environment, allowed for a framework in which I could test a lot of different AI-generating tools, while also building a story.

Here is the full video:

The constraint that AI will not always give you what you want, forces you to find creative solutions with what you get. It was easier to work with happy surprises than try and force a lesser result to your will by refining filters. The algorithm does not think like a human: the flawed aspects in the image or the footage are just ones and zeros. Creative insight is as much part of the process of making a video with AI as it is making one without.

There is an addictive reward trigger embedded in AI, comparable to receiving ‘likes” on social media. You enter a prompt, wait a few nanoseconds in anticipation, then receive the reward in the form of an image or text. It may even be more addictive than social media, just by its speed and immediacy and without potential bullying or having to please a social network.

How much time are we going to spend in front of our screens, letting algorithms create alternate realities? Observing the continuous stream of generated video on the Discord platform made me acutely conscious of all the energy it costs to run that computing capacity, and all the devices consuming it. The emissions of this process are enormous but completely invisible.

"Eveline Kolijn, a visual artist and energy activist, used AI to generate images in the style of her own work, and had similar concerns about racial and gender bias. What troubled her with the images that were produced was that AI can generate responses quickly and spawn easy, entertaining content, but little of it, for her, had depth. This was an aspect Glenn Rollans and Rosemary Griebel both identified: the (human) arts are about communicating, questioning, and inspiring, and the emptiness Kolijn noticed makes one wonder whether AI can do this. Most concerning of all, for Kolijn, was the sheer amount of energy that is consumed by the proliferation of AI content and the consumption it drives in the midst of the climate crisis."

Subscribe To Newsletter