We welcome artistic-research projects on “AI (and) Art: Poetics of Prompting.”
With recent and emerging generative AI systems, the term prompt has taken on a new meaning as a way of talking to our “intelligent” machines and instructing them to generate various content with text-to-text, text-to-image, text-to-video, or text-to-sound tools. These systems typically combine multiple AI models, such as when diffusion models for synthesizing high-resolution images converge with large language models that interpret the semantic content of our prompts, thus redefining the relationship between the sayable and the visible. Prompt engineering is developing as a discipline for refining interactions with large language models to produce “optimal” or “high-quality” responses, leading to higher predictability and standardization of the model’s behavior (we might compare prompt engineering to dog whispering, a technique of using canine forms of communication to better understand our companions but also to train them and eliminate misbehavior). On the other hand, many users mess with generative AI to surprise and amuse themselves or to play against the apparatus: prompt bricoleurs engage in a wilder form of interaction, embracing failure and serendipity.
Prompts can be understood as instructions, nudges, tasks, assignments, or recipes, and as such, they have a long and rich history within art education. Human skill and creativity have always been stimulated and trained by prompts: copy the work of your master; complete your master’s work in their style; compose a phrase using quartal harmony where each chord has at least three different pitches sounding together; paint a still life with apples in watercolor; photograph different objects to demonstrate texture… Every art school and every art discipline or medium have their own canonical, more or less formalized sets of assignments that simultaneously define what art is and what it is good for. Prompts have even become (conceptual) artworks: Yoko Ono’s book Grapefruit, Fluxus’ event scores, John Baldessari’s Assignment Sheets, or, more recently, Hans Ulrich Obrist’s Do It exhibition project are just some examples of works positioned between teaching and doing, turning routine instruction into participatory facilitation.
The prompts we invent when playing or working with synthetic media are often more interesting than the generated content. After all, the particular images, texts, or sounds generated by AI systems are the least interesting and least important aspects of these data-driven technologies of statistical aggregation. Our special issue seeks to explore the obscure realm of instructional procedurality and operationality to learn about the affordances and limitations of AI-generated media in visual arts, film, performance, music, and design, with an eye to (the tradition of) art education and instruction. The contributions should be based on practical tinkering and experimenting with AI models by challenging them with prompts, and so address the question of whether prompts are the last refuge of human creativity vis-à-vis machine learning or rather the means of optimizing and averaging our ways of thinking.
Abstracts submission deadline – October 31, 2023 to firstname.lastname@example.org, max. 250 words.
Final projects deadlines – March 31, 2024.
ArteActa is a peer-reviewed, online, open-access journal for performative arts and artistic research, offering publication in multimedia formats (text, hypertext, photo, video, audio). Project abstracts can be consulted via email at email@example.com.