17. People use artificial intelligence to generate images simply from a text description.

DALL·E is a new AI language model trained to take text-based descriptions and turn them into actual images (via Technology Review). Similar to the way we can say to you, “A brown dog that is sleeping” will conjure up a picture in your mind of a sleeping brown dog; DALL-E uses an extensive neural networking system to do the same thing (via Technology Review). It is even capable of creating anthropomorphized images. What does that mean? If you say create “an illustration of a baby daikon radish in a tutu walking a dog” then it will provide pictures of precisely that.
The AI here works by combining the text data and the image data as a single stream of data using what is called tokens. Each token is part of the program’s vocabulary (imagine each letter of the English alphabet represented as a token to get the idea) which means both text and image portions. Using this concept, the artificial intelligence can generate an image entirely from scratch or even alter an existing image to fit the required data.