Computers have learned to draw any pictures by text description

OpenAI specialists have developed two neural networks at once. One of them is capable of drawing any pictures from text

description, while the second can identify objects and classify them, relying on the description.

A neural network that was trained to draw pictures usingdescription, called DALL-E. Interestingly, the computer does not search for images on the Internet by the description, but creates them on its own. For example, this is what DALL-E gives for the query “avocado shaped chair”.