Japanese scientists developed AI that can read minds. The artificial intelligence was taught to show what a person sees before their eyes: if it is a picture with a letter, then the algorithm will draw an image similar to it.
The training process took 10 weeks. All the participants were shown various pictures, after which the scientists fixed the brain impulses they had. The impulse scanning was used to train artificial intelligence to decode information and visualize it.
Researchers observed people reacting not only during testing. Upon completion, they were asked to imagine what they saw again. The algorithm was trained to recognize these brain impulses and tried to recreate certain images.
At the first stages of training, images of geometric figures were used. The neural network was able to reconstruct even abstract objects. This lets us conclude that the program learned to generalize data on its own.