‘Paint the Picture to the Word: Shakespeare Illustration and Artificial Intelligence Art’ is a new kind of digital exhibition. I created the project because I wanted to explore and think about the potential implications AI has for Shakespeare illustration, and to give myself a better understanding of the tools that this technology offers. Every image collected here has been generated by Stable Diffusion, a powerful text-to-image AI. To create an image using this technology a user simply types a description of what they want to see into a text box and the AI will then produce several images corresponding to that initial textual prompt. If you do not like what you see, you can ask the AI to generate further images until you find one that is satisfactory. The implications of this technology are obviously profound and the questions they raise about what is art, who is an artist, the relationship between word and image, alongside various ethical concerns, are all important and complex issues when it comes to theorising and working with these tools.
AI Art problematises Walter Benjamin’s famous concept of the ‘aura’ (already destabilised by digital art) and challenges our basic ideas of authenticity and originality in the process. The images collected here did not exist before I generated them. However, they are contingent on a huge dataset of billions of pre-existing images scraped from the internet which the AI has been 'trained' on. In my experience, the technology has never once reproduced the same image, even when using the same textual prompt on many multiple occasions. The Artificial Intelligence, then, generates a unique image at the exact moment it processes a user’s textual prompt and, as such, both the human user and the AI should be understood as artistic collaborators. They create images that are products of the imagination of a user who relies on an AI that has 'learned' from the dataset to understand and interpret textual instructions. In effect, what the user, or 'Artificial Intelligence Artist', is collaborating with is a vast historical network of image-makers, both past and present.
‘Artificial Intelligence Artist’ is a term that is used frequently in this field. The artistic element in this work is found in the keywords the 'artist' inputs into the AI and the curatorial decisions that they then make in deciding what images to keep or discard. Where Hamlet’s guidance to the actors was to ‘suit the action to the word’, when it comes to AI Art, his advice would be to ‘paint the picture to the word’. Because, with this technology, words actually do paint pictures. It is down to the skill and ability of the AI Artist to understand the effect their words have on the generated images, to respond thoughtfully to them, and change their keywords accordingly. For instance, typing into the text box ‘Falstaff sat at a table’ will generate some useful results, but they won’t be particularly interesting or imaginative. A far better and more worthwhile outcome is to be more generic: ‘an old man with a white beard sat at a table’ (and to iterate on that by adding more descriptors such as ‘with a beer’, ‘with two women’ etc.), and more experimental by adding a style of art you want the image to look like such as ‘expressionism’, or ‘collage’. The permutations are endless, and the results are often fascinating and surprising.
Curatorially, then, what is presented here is just a selection of the hundreds of images the AI generated for this project. I have chosen images that are reflective of my tastes (colourful, sometimes playful, occasionally abstract), that provide a kind of aesthetic cohesion and balance despite differences in artistic styles: some of the illustrations are expressionistic (King John, Julius Caesar), while some are more literal (Merry Wives of Windsor). It was important for me, too, that the images should offer a visual idea or a gloss on the plays: Henry VIII, with the central characters represented in fuzzy felt, is grimly ironic, while in Pericles both Mariana and her father are seen through a watery prism, echoing that play’s concern with sea imagery. Twelfth Night is presented as a paper collage with a photograph of Viola and a clock – a reference to her line ‘O time, thou must untangle this, not I’ and the thematic role of time in that play. Similarly, Henry IV, another paper collage, has the sun coming out behind Hal on his horse, foregrounding his line ‘Yet herein will I imitate the sun’. There are many more such examples like this in the collection.
This exhibition also provides us with an opportunity to think about how this new technology can be applied to Shakespeare in learning environments. It is quite possible that in a relatively short period of time, educators will be asking their students to create a portfolio of images of scenes from Shakespeare, designed so that it visually represents their understanding of the plays, or to even imagine theatre or film productions that are yet to happen. Indeed, whilst this exhibition has focused on more traditional artistic practices, there is a world of photography waiting to be explored, such as these rather haunting and evocative images I quickly generated (see the bottom of the page) of scenes from The Tempest. As ever, when it comes to education, if you want to learn about something, then make it yourself.
Regardless of whether the work in this exhibition can be considered ‘art’, I hope that it provides a context and a space for us to begin to think about and engage with these powerful new creative tools and that it gives us the confidence to experiment with them.
Dr Michael John Goodman
Dr Michael John Goodman is a researcher, writer and educator who uses Shakespeare, art, and design as modes of enquiry to bring together objects and artefacts so that we may see them in new ways. He is the creator of the Victorian Illustrated Shakespeare Archive, an open-access online resource that contains over 3000 illustrations from the most significant illustrated Shakespeare editions in the Victorian period.