DALLE-dle wants you to search deep within the mechanical soul of some neural net's doodles.
Here’s a weird one: Take famous inspirational quotes, feed them to an image generation software, and see what comes out the other end. Then take the product and quiz people on which quote was used to generate it. That’s the idea behind DALL-E-dle, which gives you a suite of images, four options, and asks you to figure out which quote produced the art on display in a limited number of guesses.
You can try DALL-E-dle on its website: https://glasp.co/dalledle/
It’s very strange, honestly. Is a nebulous figure surrounded by a field of blue prompted by “You only pass through this life once, you don’t come back for an encore” or by “You miss 100 percent of the shots you never take”? Probably not either one, but it’s kinda hard to be sure.
The AI DALL-E 2 uses a process called diffusion to generate images. It starts with a pattern of random dots and a prompt. It then alters the pattern of dots towards an image as it recognizes specific aspects of that image as being similar to the prompt. It’s the second in a series of generative image programs developed at the Open AI project.
We’ve been talking about AI art all year, it seems. First we tooled around with Nvidia’s AI art program, which turns doodles into masterpieces. Then someone started using AI to generate fusions of Overwatch characters. After that, some guy won an art competition with a generative AI and then doubled down on it.
Then someone decided a person who showed up in their art was a spoopy cryptid. However, most recently, the people that make Facebook made an AI video tool and we hate it already.
About Post Author
