Advances in artificial intelligence have been progressing rapidly this century, and the technology is already widely used in our everyday lives, sometimes without us even realizing it. Those Netflix suggestions based on your likes and previous views? Powered by AI. Those incredibly annoying robotic voices you hear whenever you try and call a large company? Powered by artificial intelligence (though extremely badly, in most people’s experience). The “chatbots” that help you on websites, the “speech to text” capability of your smartphone, facial recognition technologies, self-driving vehicles, and many other things are working silently behind the scenes? All powered by AI.
Now comes a new artificial intelligence text-to-image generator called DALL-E 2, named for Spanish surrealist artist Salvador Dali and Pixar’s animated classic WALL-E. According to their website, “DALL·E 2 can create original, realistic images and art from a text description. It can combine concepts, attributes, and styles.”
In other words, you tell it what you want, and it spits out an image seconds later.
— Mirko Lorenz (@mirkolorenz) September 21, 2022
I tried the first thing that came to mind, “fairies in a strawberry field.” Here’s what I got:
Next, I figured I’d get the kids involved. “Cyberpunk samurai city at night,” one teen offered. Within seconds:
Pretty cool, if you ask me. The Washington Post has also encountered this pioneering technology; can you guess what they asked for? A photo of Jan. 6 protesters outside the Capitol, of course. What they got back was a credible “photo”—not a photo at all, but a completely artificially produced image—that was almost realistic enough to appear in a newspaper. (No, it’s not worth re-printing here.) To me what this shows is that the WP has lost any semblance of imagination, and is obsessed to the point of madness over “1/6,” but that’s a story for another day.
Back to the cool stuff. How about asking for a photo of a raven flying across the moon? No problem:
Now I’ve dabbled extensively in photography, and let me tell you, I’d be mighty proud if I’d captured that image. The technology does have its limitations—some of the things I asked for didn’t produce great results—and there are rules of course. I asked for a Picasso version of Joe Biden and was told this would violate the rules. It did allow me however to request a rendering of Ronald Reagan on the moon holding a flag, and while it faithfully outputted a pretty cool image, the face of the figure did not look like The Gipper. But try, “watercolor art of a lonely robot holding a balloon,” and you might just get a surprisingly heart-breaking portrait of android melancholy.
An “oil painting portrait of a capybara wearing medieval royal robes and an ornate crown on a dark background” also yields pleasing results. Who knew?
The Post explains the background of DALL-E 2:
Since the research lab OpenAI debuted the latest version of DALL-E in April, the AI has dazzled the public, attracting digital artists, graphic designers, early adopters, and anyone in search of online distraction. The ability to create original, sometimes accurate, and occasionally inspired images from any spur-of-the-moment phrase, like a conversational Photoshop, has startled even jaded internet users with how quickly AI has progressed.
Five months later, 1.5 million users are generating 2 million images a day. On Wednesday [September 28], OpenAI said it removed its waitlistfor DALL-E, giving anyone immediate access.
The emerging AI technologies are not without risk, and while they’re fun and sometimes useful, they’re also loaded with danger. “Deepfake” tech has made huge strides and convincing photos—and even videos—can now be produced showing people doing things they never did or saying things they never said. How long before somebody is able to produce a completely realistic video of Vladimir Putin declaring war on the U.S. or something similar? Hopefully, the good guys can stay one step ahead of the bad guys in being able to detect when photos and videos have been manipulated.
DALLE-E 2 isn’t the only game in town—before I heard about DALL-E, my son had shared a site called Midjourney which produces some wild, arresting images also based on your input. Meta and Google are also of course trying to get in on the game, and will presumably try to buy up any of these upstarts if they get the chance; after all, that’s their modus operandi. But for now, sign up for your own free account and start asking away—who knows what you’ll get. Your imagination is the only limitation. You might request a Salvador Dali painting of the Hollywood sign and get something crazy like this: