AI-augmented artworks were presented at the 29th International Symposium on Electronic Art by an artist working with Pavlos Andreadis

[04/07/2024] Yana Knight, an innovative cartoonist and illustrator affiliated with the University of Edinburgh, presented her pioneering AI-augmented artworks at the 29th International Symposium on Electronic Art. Knight's work fuses human creativity and artificial intelligence (AI) to develop new forms of non-linear visual narratives and experiences.

Photo of Yana Knight and examples of work by Yana

Yana Knight works alongside academics teaching the AI and Storytelling course. As part of the course curriculum, she leads an annual workshop titled "Creativity with AI" which uses the course library as the medium and the tool (instead of paints, brushes, and a pen; but also, alongside them).

AI as an artistic tool

The AI and Storytelling (AIST) course, designed by Pavlos Andreadis and co-developed with his PhD student Patrick Kage, encourages students to create and tell stories, using images and text. AI technology is seen as a tool that can both generate new ideas and complement those of a person, while also acting as a potential bridge for collaboration between people.

The AIST library functions as an accessible repository of a range of generative AI and other AI tools, designed to facilitate convenient programmatic access for students and artists.

Pavlos hopes that, in the future, access to the library will be extended to the public. Both the library and the accompanying notebooks were crafted with genuine creative processes in mind, drawing significant inspiration from the creative practices of both Yana and Pavlos.

AI and Storytelling is inspired by creative exercises first proposed by the renowned America cartoonist, Lynda Barry. These exercises place the creator within a structured framework, encouraging exploration of the relationship between the artist and the medium. This approach delves into the dual concept of Creative Constraints and Controlled Randomness, which in AIST includes AI as a medium.

The library provides students with tools for creating and modifying images through technologies like Stable Diffusion and DALL·E, as well as working with text — whether that's continuing a story with GPT-2/GPT-3, summarizing ideas, answering questions, filling in the blanks, or captioning photos. These capabilities make it easier for users to build their own AI-driven applications, such as chatbots or interactive art installations.

ChatGPT was launched when the course was being developed. The course material discusses this tool, but AIST encourages a different way of engaging with AI. Rather than just using AI tools in a straightforward, pre-set manner, the course encourages exploring and crafting students’ own creative methods. This approach emphasises being a creator who actively shapes content, rather than just overseeing or adjusting what the AI produces.

Yana’s journey

In her artistic practice, Yana Knight uses the course's platform and its AI library. At the beginning of the process, Yana’s original drawing is presented to the AI tool. This part is intuitive: the image needs to represent the artist’s style but provide a good basis for the generation of further images. The AI

tool generates images from the original, preserving the artist’s style, including colours, lines and composition. Prompts are used to generate the AI images.

But the relationship between the initial sketches and the final artwork is far from straightforward—it’s more of an interplay among various mediums. Yana uses the AI tool as an integral medium in her artmaking. Her artistic process is a dialogue between traditional and AI tools. Her method involves experimenting and tinkering with both the physical and digital aspects - a sketchpad, a canvas, or a digital notebook - allowing the final artwork to gradually take shape. This process involves a dynamic and continuous exchange between text and image through GenAI, alongside traditional artistic methods like sketching and printing.

The AIST tools have allowed me to easily experiment and add AI into my analogue comic-making practice and are helping me develop a combined human-AI comic-making approach. They allow me to switch between analogue and AI-generated drawings, using my own drawings as a source for AI-generated images, giving me the possibility to maintain and work around my own style and visual elements of the image (composition, shapes, colours, lines etc.)

Yana’s artworks made with the course tools have been presented at NeurIPS 2023 Creative AI expo and were on display at an exhibition at the University of Tartu Art Museum, as part of the Tartu European Capital of Culture 2024 program.

Build your own artistic AI tools

In the AI and Storytelling course, students learn the principles of machine learning and explore the strengths and limitations of AI. A variety of options that AI provides for storytelling is explored through a combination of hands-on experience, workshops on creative storytelling, and invited lectures from experts in diverse areas on AI and its application for creating stories.

As part of the course, students are given an opportunity to build AI solutions 'by hand', replicating the process Machine Learning algorithms go through.

Students are working with a high-level AI tool set that has been custom-made for the course, testing the limits of the creative potential of such tools, while also discussing the dangers of their uncritical use.

Related links