AI Art – creative or not?
In October of last year, a painting was sold at an auction for $432,500 which was 40 times the expected price (Yahoo.com, 2018). While the price tag may not be usual for classical paintings, what sets this painting apart from other paintings is that it was created with the help of Artificial Intelligence (AI).
Figure 1: Edmond de Belamy (Obvious-art.com, 2019)
The application of AI in art has increasingly been gaining traction. Mario Klingemann, a German artist, is seen as a pioneer when it comes to applying AI to create art. He says that “Humans are not original, …, we only reinvent, make connections between things we have seen” (Klingemann, n/a). While humans are limited based on what we have learned in the past, Klingemann argues that machines are not restricted by such biases – “they can create from scratch” (Klingemann, n/a). However, many would respond to this with the comment that the outcome of a machine’s algorithm is solely a reflection of the programmer’s creativity.
Now comes the part which impressed me the most when reading more about Klingemann’s work, which also falsifies the previous argument. Namely, he uses a sub-category of AI called generative adversarial networks (GANs) (Klingemann, 2019). The way it works is that the system consists of two neural networks working together. While system 1, the Generator, is creating images from scratch, system 2, the discriminator, is trained on judging the quality of those images and is capable of accepting or rejecting those images created by system 1 (Gandhi, 2018). This ultimately allows the algorithm to judge its own work, learning from it and being able to continuously improve its content.
This leads to the logical question: If creativity is a defining feature that solely applies to humans, how can an algorithm be considered creative given that it is solely a collection of wires and transistors. Well, ultimately a human brain is a collection of nerves and synapse that build a complex system. Hence, if humans can be creative, why not machines?
So, what does this mean? Well assuming that creativity is a human feature may be naïve. I do believe that machines can express degrees of creativity as they have done in the past as AlphaGo, Klingemann’s work and many other examples have demonstrated. Machines may have the potential to redefine of what it means to be creative, and as Klingemann puts it – AI may ultimately liberate us.
Sources:
Yahoo.com. (2018). Algorithm art fetches $432,500 at NY auction: Christie’s. [online] [Accessed 18 Sep. 2019].
Gandhi, R. (2018). Generative Adversarial Networks — Explained. [online] Towards Data Science. [Accessed 18 Sep. 2019].
Klingemann, M. (2019). Quasimondo | Mario Klingemann, Artist working with Code, AI and Data. [online] Quasimondo.com. [Accessed 18 Sep. 2019].
Obvious-art.com. (2019). Obvious Art. [online] Available at: https://obvious-art.com/about-us.html [Accessed 18 Sep. 2019].
