Customising Images: On Photography, Machine Learning and the Logic of the Archive

Deep-learning algorithms and models such as Dall-E, Midjourney, and Stable Diffusion that generate photorealistic images as well as more discrete applications of AI in cameras are intrinsically linked to the logic and history of photography and ‘the archive’. They operate with vast amounts of image-text, and image-metadata pairs, which form the basis of the algorithmic training processes. Especially when considering the history of stock photography, the creation of image banks and the computerization of image search starting in the 1980s, these technologies remediate patterns of standardization, generic aesthetics and the deflation of the semantic layers of images. The talk aims at situating a selection of current developments in digital imaging within that history and argues that the generation of future images resides in concepts of the image as a malleable and exploitable resource.

• Æstetisk Seminar E-2023 Program
• Æstetisk Seminar E-2023 Plakat 

Æstetisk Seminar efteråret 2023 er tilrettelagt af Nicolas Malevé og Jacob Lund, med hjælp fra Annelis Kuhlmann og Jussi Parikka på vegne af 
Institut for Kommunikation og Kultur, Aarhus Universitet.