This event will not be the deal of the century, but it will definitely go down in the history of the development of intelligent systems. At the prestigious London auction Christie's officially sold the first art painting, written by a neural network. More precisely, a group of French creatives called “Obvious” showed an example of how to make substantial money on a HYIP topic using program code.
Obvious took an open source neural network that was developed by someone named Robert Barrat. The enthusiastic programmer is only 19 years old and has posted his "pen test" on Github for peer review. The neural network was not meant for creativity, it is just a set of interconnected algorithms, but the French adventurers liked it.
The stake was made on the fact that for most people the concept of artificial intelligence is abstract, they do not distinguish a machine learning system from adaptive tuning, etc. And therefore, even at the stage of "drawing", Obvious launched something like an advertisement, they say, we are training a new outstanding virtual artist. Real creativity from the car! In fact, the designers routinely fed the neural networks with examples of real pictures from Wiki Commons, trying to get it to start drawing something similar.
The specialists of the auction house agreed to recognize the final version as a "painting", and the accumulated interest helped to sell this drawing for almost half a million dollars. Everything is fair and the Obvious group even officially expressed gratitude to the author of the neural network, Barrat, although they are in no hurry to share money. They are in their right - no laws are violated and anyone can try to get their "masterpiece" from the neural network, fortunately, it has remained in the public domain.