To say that Artificial Intelligence is reaching limits unimaginable just a couple of years ago is nothing new. Through deep learning, artists and computer scientists are teaching machines to create. And with excellent results. It is becoming increasingly common for machine-painted pictures to be bought for incredible amounts of money. They are also winning prizes that are highly valued in the art world.
This is the example of Mario Klingemann. The German artist has created an algorithm who has "painted" the work called 'The Butcher's son' and who has won the highest award in one of Europe's best-known alternative competitions, the Lumen Awards. In this case, Klingemann taught his algorithm with classical paintings and pornographic images and videos.
Anna Ridler, a London-based artist, also teaches machines to draw. For years, she has been interested in drawing and painting, but also in technology, databases and big data. Ridler started to use machine learning and to make their computers to create on their own according to what they learnt. So she began to use machine learning and let her computers create. As the artist explains in The EconomistWhen you talk about machine learning, the decisions are no longer made for a particular drawing, it's more about the decisions you make when you create your dataset, what you leave in and what you take out.
AICAN is another such example. The algorithm has been created by the art and artificial intelligence laboratory at Rutgers University (New Jersey) and already sells paintings for a whopping $16,000. Without knowing that they have been painted by an Artificial Intelligence, of course. His drawings are so good, that 75% of the time, people thought that the images created by AICAN were created by a human.
More than 80,000 different images representing the artistic canon of the last 500 years were used to teach AICAN to paint. In this way, he was not "educated" in any particular style and could be "free" to create.
The Obvious collective, made up of three young Frenchmen, has also made art history thanks to algorithms. They managed to sell one of their paintings at Christie's New York headquarters for $432,500. As Vernier, one of the members of the group, explains, in a article in El PaísWe started working on the painting nine months ago, when we came across a series of algorithms that surprised us with their possibilities. We felt that creating a work of art was the best way to demonstrate what an algorithm is capable of.