The horse, astronaut, and stars depicted here do not exist in reality. They are the imaginative creations of a computer model called DALL-E 2, developed by OpenAI in April. This model, a blend of surrealist artist Salvador Dalí and Pixar’s robot WALL-E, learned to connect words and images by analyzing a vast database containing hundreds of millions of pictures and their corresponding descriptive labels. When given a simple phrase like “a photo of an astronaut riding a horse,” DALL-E 2 generates an image based on its comprehension of the terms “astronaut,” “riding,” and “horse.” It even adds related contextual details, such as placing astronauts against a starry background, drawing from its ability to associate connected concepts.
