We've seen plenty of AI image generators in our time, and while some have impressed us, there's always been that uncanny valley quality to their output. But right now, the internet is going absolutely wild for a new text-to-art generator – and it's easy to see why.
OpenAI's DALL·E 2 system is able to transform a simple sentence into a fully-fledged piece of art. And it seems the only limit is the user's imagination – from dinosaurs playing frisbee to horse-riding astronauts in space, you name it, and DALL·E 2 will make it. See our piece on how to use DALL·E 2 for more details on how it works.
A "portmanteau of the artist Salvador Dalí and Pixar’s WALL·E", DALL·E 2 is already provoking a mixture of awe and terror online. There's something pretty incredible about how accurately the tool is able to turn text prompts into images. Take a look at some examples below and you'll see what we mean.
So how does it work? According to OpenAI, a process called "diffusion" starts with a pattern of random dots and gradually alters that pattern towards an image when it recognises specific aspects of that image. Through this process, DALL·E 2 has "learned the relationship between images and the text used to describe them."
But, in the words of [insert Spider-Man character], with great power comes great responsibility – and it seems OpenAI is taking steps to mitigate the potential of the tool being put to ill-use. Not only is the tool unavailable to the public (though you can join a waitlist), but it is also unable to generate real faces, or create NSFW images.
Still, even with these safeguards in place, the power of the tool is causing some concern online – mainly with regards to what it means for artists. "Not gonna lie, didn't expect "illustrators" to be one of the first human jobs that artificial intelligence completely displaces but here we are," one user tweets, while another adds, "The implication of AI replacing artists was always there, and now the day has unfortunately come."
We wouldn't go that far – while the tool is impressive, the results are still pretty random. But like the most terrifying deepfakes, it's the potential for misinformation that's scary – AI content creation clearly isn't going anywhere, and the results are only becoming more impressive.