OpenAI trained the software, known as Dall-E, to generate images from short text captions. It specifically used a dataset of 12 billion images and their captions, which were found on the internet.
The lab said Dall-E — a portmanteau of Spanish surrealist artist Salvador Dali and Wall-E, a small animated robot from the Pixar movie of the same name — had learned how to create images for a wide range of concepts.
OpenAI showed off some of the results in a blog post published on Tuesday. "We've found that it [Dall-E] has a diverse set of capabilities, including creating anthropomorphized versions of animals and objects, combining unrelated concepts in plausible ways, rendering text, and applying transformations to existing images," the company wrote.