Ah, AI art. The technology has evolved so quickly that it’s hard to believe it’s only been 12 months since we started talking about the latest generation of text-to-image generators. No doubt, another year will see things progress even further and some of the strange artifacts appearing in AI art will be resolved. But for now, AI image generators still create some pretty weird pieces.
If you’re interested in working with an AI image generator yourself, check out our collection of the best AI art tutorials. In the meantime, as we approach his one-year anniversary for diffusion models like DALLE-2, Stable Diffusion, and Midjourney, we round up the weirdest AI art we’ve seen so far this year.
01. The Pope in a coat
One of the most terrifyingly compelling AI-generated “pictures” of celebrities we’ve seen has to be the AI Pope in a coat. Created using Midjourney, this image took the internet by storm and provided a perfect example of how good image generators are at replicating celebrity likenesses. I didn’t look too closely to realize that even the AI images of many photographed celebrities are quite off and they aren’t real. However, this seemed so perfect that many people fell in love with it and raised concerns about how to tell facts from online fiction.
02. A woman eating a salad
One area of photography where AI-generated imagery can have a significant impact is stock imagery. We’ve already seen some of the best stock photo libraries start to incorporate AI, and some brands are using AI-generated “models.” A classic trope for stock images was the “woman eating salad” genre. So how well can the best of her AI image generators handle that as a prompt, which turns out to be pretty scary without serious tweaking. A Twitter user shared a series of AI images created via a text-to-image prompt, “Woman laughing with a salad,” creating an entirely new type of fear in the process.
03. Will Smith eating spaghetti
What I enjoy most about working with AI is the sheer unpredictability 😅 Surprise of the day: this text-to-video generated spaghetti-eating Will Smith monster 🍝 pic.twitter.com/ fTOY4LBTRdMarch 28, 2023
Compelling AI-generated video seems a long way off. Generating a moving image is much more complex than a still image, as video by definition requires multiple frames and requires consistent composition and lighting. For now, AI videos tend to look glitchy and contain a lot of flicker, and some experimentation is no less disturbing.For this AI video of him of Will Smith eating spaghetti, Sam Asante said: It goes far beyond how I’m going to eat spaghetti if I order it on a first date.
05. AI Live Action South Park
Here’s another AI video. This time, we bring cartoon characters to life in the live-action South Park. The characters look frighteningly real…almost. Created by YouTuber demonflyingfox (opens in new tab) Using Midjourney plus animation software, this video has the color and quality to make it look like it was actually made in 1997 when South Park came out. It creates a deep sense of unease. All that silent blinking is very disturbing.
Want to see more? For more bizarre gems, check out our selection of bizarre AI art.
