Visual Epistemology and Artificial Intelligence

Asst. Prof. Dr. Fabian Offert

In 2024, two Nobel prizes were awarded to artificial intelligence researchers, and artificial intelligence models are now indispensable to a wide variety of scientific endeavors, from analyzing distant galaxies to modeling novel proteins. We are witnessing the emergence of a new artificial intelligence trading zone – a zone of local collaboration despite global difference, where a method reshapes how knowledge is produced. The social, political and cultural implications of this trading zone, however, have so far not been studied in depth. Crucially, models trained on cultural data, especially large multimodal models, now routinely inform scientific experiments and scientific reasoning. The research project – which is intended to lay the groundwork for my second book – thus asks: how can theories of the image, from art and art history, inform a close-reading of the artificial intelligence revolution in the sciences? Can we imagine an "aesthetics" of scientific reasoning, now that almost all data is cultural data? And does the universality of artificial intelligence models suggest a convergence between humanistic and scientific forms of knowledge-making, finally bridging the "two cultures"?