DALL-E, make me another Picasso, please

Ever since humans invented art, sometime in the Paleolithic era, they’ve taken a lot of pictures – “The Starry Night,” some memes, that picture of Donald Trump staring at the solar eclipse. What does it all add? A few years ago, a company called OpenAI fed a large portion of those images, along with text descriptions, into the neural network of an artificial intelligence called DALL-EDALL-E was trained to create original art, in any style, depicting almost anything in uncanny detail, based on written directions. But mastering the entire universe of human images makes for difficult choices. How do you determine what? DALL-E need to create? After careful consideration, one of the first images OpenAI conjured up was a donut made from porcupine quills.

“There was a belief that creativity is something very special, just a human being,” OpenAI CEO Sam Altman recently explained. Maybe not so true anymore, he said. Altman, who wore a gray sweater and had disheveled brown hair, was videoconferencing from the company’s headquarters in San Francisco. DALL-E is still in a testing phase. So far, OpenAI has granted access to a select group of people – researchers, artists, developers – who have used it to produce a wide variety of images: photorealistic animals, bizarre mashups, punny collages. Asked by a user to generate “a plate with various alien fruits from another planet, photo”, DALL-E returned something like rambutans. †The rest of Mona Lisa” is, according to DALL-E, usually just one big cliff. Altman described: DALL-E as “an extension of your own creativity.”

For the more than a million people on DALL-E‘s waiting list, the only way to increase their creativity is to slide a request to the AI’s Instagram DMs. The company launched the account, @openaidalle, in April. “I was concerned that maybe more explanation would be needed to get people involved,” said Natalie Summers, who manages the account for OpenAI, from a conference room near Altman. “And that didn’t happen.” @openaidalle has almost two hundred thousand followers. Summers’ job is to read through the posts and pick the best of the best. “If I did everything people asked, we’d have a lot of raccoons and sloths,” she said. Hits included “cheeseburger lamp”, “emotional baggage” (suitcases with sad faces) and “attractive dinosaur in tuxedo, looking at itself in a mirror and seeing its reflection, digital art”, which DALL-E endowed with human proportioned arms to aid in his hotness. Reviews are of the mind-boggling variety. “I’m going to lose my job,” one commenter, whose profile said they were a graphic designer, posted under an image of polymer clay dragons eating pizza on a boat.

To review the latest requests, Summers, who wore dangling earrings and a denim jacket, videoconferenced with DALL-E‘s product manager, Joanne Jang, and with a member of the technical staff, Aditya Ramesh. Ramesh was responsible for DALL-Ethe name; it came to him in the shower. “Some people got it right away,” he says. “I had to explain to other people that it’s a contraction.”

There are rules for applicants. Depictions of public figures are off limits, as is anything remotely offensive, including nudity and violence. Conducting political campaigns is prohibited. “We are concerned about deepfakes‘ says Summers. Recently, a researcher named Boris Dayma developed a low-fi copycat called DALL-E Mini, which went viral. Users were allowed to submit prompts like ‘Ice T in a glass of ice tea’ and ‘Babies fist fight’, although the output is sometimes creepy: Ice-T’s face seems to melt; the babies look like zombies. At the request of OpenAI, DALL-E Mini was renamed Craiyon. (“There has been a lot of confusion,” said an OpenAI spokesperson.)

Summers started scrolling. “Here’s one I found this morning: ‘a cat with a bed of tulips growing out of its back,'” she said. She clicked a button and the system came up with ten images. All featured tulips, but only one had tulips growing out of a cat’s back – though not enough to qualify as ‘a bed’. Ramesh tried the prompt, and the machine spat out a white cat with about twenty tulips sprouting from its fur. “I love how chubby the cat is,” Jang said. Summers sent the photo to the user, who responded with three smiley emojis.

They turned to “an astronaut eating in a restaurant floating in space.” Jang saw a problem with DALL-E‘s work: too much gravity.

“Oh, that’s true,” Ramesh said. “I’ll see if I can get floating food.” A few tweaks to the wording resulted in an astronaut with a piece of toast, gazing at the stars out of a restaurant window. “It’s like he’s thinking about the decisions of his life,” Ramesh said. They went for a less melancholic option.

Next: fishing a fish. “How about this one?” said Ramesh, drawing up a picture of a green fish in a bucket hat, with a smaller fish dangling from a rod.

“Which is good because he looks alarmed to be in this meta-situation,” Summers said.

The meeting drew to a close, but they decided to consider a few more requests. A user simply asked for “The Big Bang”. Jang took a deep breath. “There’s a lot of artistic freedom for that,” she said.

Leave a Comment

Your email address will not be published. Required fields are marked *