In 2018, a work of art generated by AI (artificial intelligence) was sold at Christie’s, a first in the 252-year history of the auction house, for 432,500 US dollars, nearly 45 times its initial estimate of 7,000 to 10,000 US dollars. A representation of a fictional character, the work titled ‘Portrait of Edmond Belamy’ was created by Obvious Art, a Paris-based Collective of artists, using a machine-learning algorithm of GAN (Generative Adversarial Network) type. The tool is an algorithm of Ian Goodfollow (Google Brain). The system was fed with a data set of 15,000 portraits, painted from the 14th century to the 20th century. The 70 x 70cm portrait (and others representing the character’s ‘family’) was printed and put into a classically inspired frame. The signature on the portrait is an algebraic formula, min G max D Ex[log(D(x))]+Ez[log(1-D(G(z))].
Created with robotic or automated technology, robotic art exhibitions featuring the work of robotics artists from around the world have been organised and shown by Art Bots since 2002. Fast forwarding to June 2019, a solo exhibition of drawings, paintings, sculptures and video art, called Unsecured Futures and created by a humanoid AI robot called Ai-Da, opened at Oxford University. Ai-Da was designed by a Cornish robotics company called Engineered Arts, while engineers in Leeds developed the robotic hand and pencil to draw what it sees with a camera in its eye.
As a new kid on the block, Robotic Art is raising alarm bells in the conventional art world. Can machines ever create art that moves us? Can it truly be described as ‘creative’ or ‘imaginative'? And what is its potential?
Robotic Art, the creative art world’s most challenging new frontier, has begun to evolve rapidly
These are some profound and fascinating problems that could redefine the mysterious possibilities of human creativity. Practical aspects, such as provenance, legal authorship, sales distributions and copyright and ownership issues also need a new set of functional SOPs. The Belamy print became controversial when Obvious Art admitted to using code from another AI artist, 19-year-old Robbie Barrat. It is also not clear if Barrat can make a claim for ownership of the artwork as his code was shared under an open-source license.
Going beyond conventional painters, sculptors or printmakers, Robotic Art is the handiwork of individuals, groups or companies that are essentially technology specialists.
Going beyond conventional painters, sculptors or printmakers, Robotic Art is the handiwork of individuals, groups or companies that are essentially technology specialists. This combination of art and engineering advances both fields — but machines don’t make art. Robots that paint operate in two general ways — either manually or through telekinetic remote operation of the robot arm and, secondly, by getting painting commands that are created by software. For robots to paint with direct human involvement, a physical tool that a human artist can move and have a robot mimic is used. The artist’s remote motions function in a manner similar to how remote doctors can perform surgery.
As there are still no clear rules in technology about creating physical art, the actual painting process can be unpredictable. But AI is making machines more creative, which will enable artists to attempt art that is perhaps intellectually or physically more ambitious than before. Future AI advances in human mimicry or extension might also affect the fundamental connection between the artwork and those who interact with it. However, programmed machines lack their own creative will and, for the moment, it is still the human artist as visionary who decides how to artfully use these new tools.
Published in Dawn, EOS, August 16th, 2020