The transmitter sends only a rough sketch of the message — like an artist drawing contours without details. At the receiving end, the AI, like an experienced restorer, reconstructs the full picture, cleaning it of interference. This is how LiTCom technology works, changing the game in future 6G networks. In traditional communication, as in the standard model, the transmitter does all the work: encoding, protecting, combating noise. Here it is nearly idle — only slightly smoothing the signal, as if blurring the contours of a sketch. Radio communication relies on Maxwell's equations, but LiTCom turns their application upside down, making the transmitter passive. The entire burden falls on the receiver with a generative AI, which, from a mixture of signal and noise, as if from muddy water, extracts the original data. Chaotic noise increases entropy (a measure of disorder), but the neural network is trained to find order even in chaos — just as the ear picks out a familiar melody from a cacophony. Result: transmitter load drops by 95%, and smart amplification of important bits makes the connection more reliable. Miniature sensors with a button-cell battery will be able to send data for years from remote locations.
The concept of entropy came to communication theory from Boltzmann's thermodynamics. Today it underpins the reliability of the internet, but its true nature is inevitable disorder, which, it turns out, can be turned to one's advantage.
AI that completes images from noise is not science fiction but reality, where training on megadata replaced the 'magical enhancement' from spy movies.