The Google DeepMind division, engaged in developments in the field of artificial intelligence, taught the computer "mind" to generate photorealistic images based on text entered by a person. As reported by Futurism, search engine algorithms create a completely new picture, not similar to what they "saw" before.

This is not the first Google experiment in the field of creating images using AI. In February, the team of the Google Brain research project, focused on in-depth training, developed an algorithm capable of "recreating" images, having on its hands a source of only 64 pixels.

Google's neural network actually recreates the image from scratch, starting from the many previously seen samples.

Based on the same technology, AI has learned to turn text into pictures. According to the researchers, the more detailed the user will describe the query, the better the result (for example, the query "yellow bird with a black head, orange eyes and an orange beak" will return a picture of a very high quality). How the algorithm works, shown in video.