AI-Powered Creativity and Radically New Possibilities in Art and Science
Abstract Contemporary artificial intelligence (AI) advances to spheres that have for a long time been considered the last refuge of human creativity: art and science. A decade ago, machines that can write short stories, produce pictures that win art competitions, or solve scientific problems were nothing but a matter of science fiction—today, they are a matter of fact. This demonstrates that the possibilities of what AI is capable of has radically widened over the last couple of years. However, the way AI is currently used for artistic and scientific purposes exhibits limitations. Text-to-image models, for example, imitate or synthesise visually representable features and styles based on the statistically salient features they have extracted from the image datasets they have been trained on. Consequently, they are largely restricted to artificial imitation, so to speak. At the same time, in the sciences, AI is mainly used to accelerate scientific processes. Therefore, scientists mainly use AI for artificial acceleration, as it were. This shows that the way AI is currently employed in artistic and scientific contexts opens up new, but not radically new, possibilities.
The proposed keynote explores AI’s potential for opening up radically new artistic and scientific possibilities. To that effect, I sketch what I call the “possibility profiles” of the domains of art and science. Possibility profiles chart the structure of viable (or “safe”) possibilities within a certain domain, i.e., possibilities that represent meaningful and valuable contributions to the respective field, without radically questioning the status quo or transforming it. Since different things are considered meaningful and valuable in art and science, the possibility profiles of both domains vary considerably. I demonstrate that the predominant modes of using AI in these fields—artificial imitation and artificial acceleration—as such prove insufficient to productively move beyond viable possibilities and, thus, to change the respective possibility profiles. The keynote examines AI’s potential for creating radically new possibilities, rather than only viable possibilities, in art and science that are capable of substantially transforming the possibility profiles of these domains. In doing so, the talk casts light on the creative possibilities and limitations of AI. Following the spirit of Possibility Studies, the proposed keynote treats “[t]echnological developments[…] as windows towards possible futures” (Glăveanu 2023: 5) by exploring future artistic and scientific possibilities through the window of AI.
Jonas Bozenhard
Dr. Jonas Bozenhard is a postdoctoral researcher at the Institute for Ethics in Technology at Hamburg University of Technology. In 2023, he completed his DPhil (PhD) at the University of Oxford with a thesis on creativity and artificial intelligence (AI). His research addresses issues in the philosophy of technology and creativity, AI ethics, value sensitive design, aesthetics, and AI-powered innovation. Additionally, he has experience in advising and working with AI-related research groups, initiatives, and companies across more than five countries. He is also passionate about communicating current debates in AI ethics to general audiences and developing new teaching formats at the intersection of creativity, responsible innovation, and AI.