I’m completely addicted to the new artificial intelligence art app Dream, which allows users to create algorithmically generated images with the click of a button. It has been used to see how AI recreates American symbols and history, such as the stars-and-stripes designs above, which were created using the term “United States of America.”
To use the app, enter a term, select one of 11 art styles (such as steampunk, synth wave, or fantasy art), and tap “create.” The images appear in seconds, as if compiled from the average of all Google Image searches for the term, though WOMBO, the Canadian AI company that created the app, does not specify how the images are created.
To find the best image recipes, the app relies on experimentation and repetition. More specific terms can sometimes produce better images, such as the images of the March on Washington, which became more recognizable as I added more words. Using “Pearl Harbor,” a horrifying combination of the memorial and images from the 1941 attack was created.
Historic figures and celebrities are one of the most enjoyable categories to create images for. Using the festive filter, I generated images of our first president in the snow, while “Abe Lincoln” displayed Lincoln’s stovepipe hat against a rock-like background, which I assume came from all of the images of the Lincoln Memorial and Mount Rushmore.
It has even attempted to recreate presidential campaign logos, yielding portraits of former Presidents as half-men, half-brands, such as former President Barack Obama as a mash-up of Sol Sender’s O logo and Shepard Fairey’s Hope, or former President Donald Trump’s hair on a glitched-out word mark.
AI has advanced since Google’s “demon puppies” days, and Dream raises concerns about the future of art if anyone can create stunning works with the click of a button.