Abstract
Generative Adversarial Networks (GANs) were created in 2014 by Ian Goodfellow and colleagues. GANs are AI algorithms that use two neural networks (the generator and discriminator) competing against one another to generate new synthetic data. By employing not one, but two distinct neural networks, GANs have produced astounding results that were previously thought to be unattainable for artificial systems, such as the capacity to create fake images of realistic quality and transform a scribble into a photograph-like image or designs, without the need for enormous quantities of training data.
When GANS are used in the fashion industry, where the role of the image is central, a number of ethical, political, and social questions emerge. GANs can, for example, be used to generate fake images, which poses a huge threat to the privacy of individuals. As a result, their use has already been made illegal in some places. Although investigating the ethical implications of using GANs is of great import, this dissertation asks a different question. This dissertation asks whether GANs are creative, using examples of their use in the fashion industry to develop my position.
I provide an exposition of the selected philosophical positions on creativity in this dissertation. While some theorists argue that GANs are not creative because we do not understand what creativity means, others argue that even though we do understand what creativity means, GANs are still not creative because they do not fit into their understanding of creativity. Another set of theorists contend that creativity is all about being autonomous, and since GANs are autonomous programmes they are creative.
By closely analysing the work of three of the most significant theorists on creativity - Margaret Boden, Berys Gaut, and Harold Cohen - I develop a nuanced account of creativity, and using that account, argue that GANs are creative in their use in the fashion industry. My position is, however, that their kind of creativity is not akin to human creativity because they lack an essential attribute of human creativity - the first-person experience of the world. This first-person experience gives us a unique and distinct style of creating, which GANs do not have. Instead, I propose that GANs’ creativity could be understood in terms of what I will call “collaborative creativity”, meaning that they make a generative contribution to the creative process.