We've been making machines that perform automatic art for centuries. The earliest known example of a clockwork device is from the first century BC in which a Greek produced astronomical chart depicts the heavens by the time of year. In China, the famous engineer Yan Shi is written to have produced an automatic clockwork human being in the 3rd century BC. This item however is lost to time. In the 14th century, clockwork became widespread and many devices were produced there-after. This included music boxes and theatrical devices, bringing art to machine in a proper way. Self-playing pianos and kaleidoscopes appeared soon after this with mechanical art generation becoming something we grew to expect from certain devices.


In this modern era, the computer is the power machine that has overtaken the engine in modern productivity. With the digital electronic circuit we can calculate numbers at phenomenal speeds and when we allow the numbers to represent other things, we can program the computer to calculate real world or imaginary situations. When looking at the world with a quantitative mind, and we assign numerical values to everything, how can we translate this into art? Even paint by numbers has been designed by a human and the instructions are given by an artist.


There has been a sudden growth of AI art generation in recent years. Many tech companies are investing serious energy into creating futuristic software that produces art at the push of button or two. The main differing factor on the user side is growing into two distinctive factions. One is dependent on user interaction and one is not. The art we interact with in order to create can ask us for input such as style, influence, colour choices, and any other variable we can think of. Our input directly affects what is produced. The other kind of AI art generation is the kind that works on its own merit. The program is allowed to run itself and produce whatever output the code informs. When AI generates intelligent art, it must draw upon known principles and ideas that it has learned from the input of others. It either learns these itself with a process called machine learning or it is taught by programmers who choose what to put in themselves. In either case, the source material for the machine's art generating code is chosen by human hands at some point. Perhaps when several AI generated databases exist, we can ask AI to teach AI to further distance the human influence.


Artists have been in two minds about the benefit of AI art generation. Clearly, as an artist myself, I can understand that it is a little emasculating to be told a machine can do your job. Perhaps it can do it better than me? You can be the judge of that. But we know a machine or a computer has no ability to feel. Not just a hungry stomach or a cut knee but they can't feel hope and despair, love and disgust, positive or negative. How can we relate empathically to a device that is simply doing what it has been told to do?


For centuries we have admired the clouds and the stars. The images created by the forces of nature playing out in random and governed ways have given us cause for stories, celebration, and song. The mythology of the night sky and the endless shapes we envisage in the movements of billowing clouds have been timeless sources of joy for all of us. No human hand produced these things, and even if we believe that the laws of nature are somehow programmed or allowed to be how they are by a presence like our own, the end result is as mechanical as clockwork. Only at the quantum level, beyond the proton, can we begin to say truly random events occur. Our minds may be some kind of quantum computer but AI programs so far all work on strict and rigid instructions. In this way, the art of AI systems can be likened to clouds or star formation, in that the results look beautiful even if we know they weren't made on purpose.


Do you want to learn about making your own AI art? Here's a great online course that'll put you aeons ahead of your time.