Artificial intelligence is gaining an increased place in our society, by optimizing our processes and changing our interactions. Other algorithms have emerged, offering some impressive capacities in terms of replicating other processes which happen in our brain. Not only are we able to automate simple logic. As an exemple, Generative Adversarial Networks are algorithms which allow the creation of entirely new and unique pictures, based on a large number of exemples. These algorithms are the first to allow the expression of some creativity by a machine. In October 2018, Edmond de Belamy, a classical portrait created using those techniques by the art collective Obvious was sold at Christie’s for nearly half a million dollars. Since then, many artists have started exploring the capabilities and limitations of such tools.
The GANs algorithms are composed of two parts: a generator and a discriminator. While the generator tries to create new images which look like the ones of the initial database, the discriminator tries to determine whether the images showed to it come from the generator, or the database (ie. if they were created by humans, or by a machine). The algorithms train until the discriminator can’t make a difference. As a result, we obtain a large number of images which are entirely new, although they fit the features of the images of the database.
By nature, those algorithms constitute an incredible tool for assisting in visual creation. They have been used on our cultural heritage, but there are many other applications to be found for those algorithms, in every task involving creation. We can expect to see the use of AI increase greatly in industries such as design, architecture, music and written content creation.
Obvious’ participation in the Parisian edition of the Nike by You workshop is a first step in demonstrating this point. They were selected as an artist for creating Nike sneakers, which will be made available on Nike.fr website. They decided to let the algorithms create the design, by training them on a large number of images of the Nikes which were offered as a basis for customization: air max 1, air max 90, and air max 97. The algorithms created an infinity of new colors for those models, and one of them was selected to be used for the creation of the new sneakers, making it a reinterpretation of the heritage created by Nike.
If we push the experiment further, we can use those algorithms on a larger variety of shoe models. As a result, the algorithms would create models which are entirely new in their design and their colors. This observation is extendable to any artistic creation from designing a chair to a building. The possibilities brought by this tool are limitless, as long as the models have a sufficient amount of data to train on. Considering most historical brands are digitally keeping track of their past creations, these tools offer an interesting way of exploring the heritage of a brand, and reinventing it.
Moreover, these new artistic tools give an access to creation to people with a different set of skills, which can only be beneficial as the creative industries are in constant search of different and innovative ideas.
Obvious’ Nike will be available as a limited edition on Nike.fr from the 19th until the 28th of June 2020. For more information, visit their instagram @obvious_art.