Over Christmas break, I spent a good amount of my free time making art. I don’t mean with a paintbrush or some colored pencils, but with an app on my phone called Dream. Created by Canadian company WOMBO (known better for its lip-synching deepfakes), Dream lets you “create beautiful artwork using the power of AI.” Enter an idea using 100 characters or less, select an art style and the app will create a picture based on the prompt in seconds.
Dream first appeared on the App Store in late 2021, and social media has since become obsessed with it. There are TikToks where users ask the app to create images based on book titles, and some creators have made a “game” in which they enter celebrities’ names into the app and have viewers guess who it is based on the likeness. Some Twitter users have even created accounts specifically for displaying Dream art, including this one dedicated to Taylor Swift lyrics (a personal favorite). The pictures circulating online are beautifully bizarre, yet somehow encapsulate their subjects.
Dream is certainly not the first of its kind; AI art has been around for years. A portrait made using GAN technology sold for $432,000 in 2018. Many professionals and critics are not worried about AI replacing traditional art anytime soon, as most computer-generated pieces are simply copying patterns and structures that already exist. Instead, AI art’s strength lies in human collaboration. Take South Korean artist Domin, who joined forces with an AI artist as part of an exhibition at the AI.A Art Gallery. Domin designed half of a scene of Dokdo Island — once with pen, the other a painting — and the AI completed the picture. Both designs were then connected, combining not only Eastern and Western art styles, but also human art with computer imagery.
From both a technological and an artistic standpoint, Dream is a way to make art more accessible. It’s incredible to think that all of this work is done using algorithms (WOMBO has been quiet on the specifics of how the app works, though it is believed to use VQGAN+CLIP technology.) I don’t pretend to have any more than a basic understanding of how AI works, but that doesn’t affect my admiration of such an accomplishment. The process of developing and implementing AI from start to finish can cost up to $1,000,000, yet Dream is free to use (although purchasing larger prints of your artwork is an option through the app.)
To put it plainly, not everyone has a natural talent for traditional art. While typing in a few words and letting software do the rest is in no way comparable to the hours of time and energy spent on a detailed canvas, this medium broadens horizons and allows for anyone interested in digital art to experiment.
One of the reasons that I love the app is because of the level of complexity it offers. In movies or music compositions, creators have a talent for telling a story with the tiniest possible details: a slight change in facial expression or a building crescendo. In physical pieces of artwork I find it harder to pick up on those minutiae, at least at first glance. The artwork created with Dream manages to perfectly capture complex ideas and emotions in a more visible way — for example, I ran the title of this article into the app in order to create the cover image. Somehow the algorithms have managed to capture this idea perfectly: the beauty that can come from going digital. It’s fascinating to see something so complicated come to life so quickly.
AI art may not surpass human talent anytime soon, but it is certainly here to stay. Inventions like Dream make the technology available to the public and provide a greater appreciation for art as a whole.
Daily Arts Writer Hannah Carapellotti can be reached at hmcarp@umich.edu.