DALL·E is trained a GPT-3 neural network by OpenAI (the makers of GPT-3) that creates images from text captions for a wide range of concepts expressible in natural language.

For examples, see the gallery, where you’ll see what DALL·E creates for the following phrases:

• 🥑 an armchair in the shape of an avocado
• 🕦🟢 a pentagonal green clock
• 🦔 a cube made of porcupine

You can even play with the examples on the DALL·E example page – give it a go.