Abstract by AI: Augmented emotions in machine learning art
I present the use of generative adversarial neural networks as mechanisms for stimulating emotions in viewers, using the project Abstract by AI as an example.
The history of painting began as drawing natural objects, and modern painting has entered the era of abstract painting. Nowadays, AI technologies are increasingly making their way into the realms of art and design. StyleGAN, in particular, is popular for creative expression as it allows users to produce artworks generated on huge amounts of art paintings and images. However, this new technology is often used to recreate images in popular art styles. Rather than reproducing a specific visual outcome based on one artist’s work, I use and explore this technology to generate new visual forms, giving new meaning to the images generated and thereby evoking emotion in the viewer.
Introduction
The use of Generative Adversarial Networks (GANs) is quickly becoming a popular means of entertainment and open-ended artistic expression and experimentation by recreating expressionist paintings or copying the styles of well-known artists. However, important art is frequently seen as a uniquely human accomplishment. To make artwork that may evoke profound feelings and emotions, the viewers perception widely assumes a human artist must possess a unique set of talents, knowledge, and genuine intent.
In this project, I am investigating whether we can use computer-generated artworks to evoke emotions in the viewer and give new meaning to the images by exploiting the possibilities provided by new technology, and use it as a tool to create new and unique forms of abstract artworks rather than imitate existing ones.
The process
Using StyleGAN, I generated abstract images out of 300 collected pictures of various styles, pictures and paintings of abstract art works. The resulting model was used to generate unique abstract images. As mentioned, the goal was not to create images to replicate a specific style of abstract art, but rather focussing on the mood the outcome might trigger in the viewer. This stands in contrast to other artistic approaches such as creating new impressionist paintings [1] or art based on Picasso and other famous artists [2]. To achieve my goal, I collected small scale number of various styles and let them train for several ours.
Designers perspective on art: Form follows emotion.
Exploring technology as a designer with a functional background, I was identifying opportunities within the ever-changing landscape of emerging technologies that can be leveraged for more experimental approaches. Having a designer’s perspective influenced how I approached the project from the start. While I was pondering about technology and the designers’ approach of putting people at the center of design, I wondered how such technologies could affect people in other ways, rather than on their actual behaviour, as in user–centered design approaches.
Emotions as a response to machine-made abstract images
According to participant feedback, the created visuals can provide participants with surprise, while also leaving them with the somewhat paradoxical and mixed feelings of having an emotional reaction to the artworks despite knowing they were created by a machine.
Conclusion
With my self-initiated work Abstract by AI, I aim to highlight opportunities in the ever-changing landscape of new technologies to provide surprising, evocative impulses, stimulate the imagination, and evoke emotion. Rather than arguing that new technologies are upending the art world by replacing human-created art with machine-created art, I see these technologies as an opportunity to expand the field and serve as a playground for exciting new ways of creating art.
[1] R. Gonsalves, GANscapes: Using AI to Create New Impressionist Paintings (2021), Towards Data Science via Medium
[2] E. Packard, The Intersection of Art and AI: Identifying and Generating Famous Works using CNNs and GANs (2019), Becoming Human: Artificial Intelligence Magazine via Medium
