2022 saw an explosion of text to image synthesis Artificial Intelligence (AI) models released into public domain on the web. Vastly outperforming prior implementations of ‘computational creativity’, these models are best understood as enabling users to explore boundaries within latent space (the compressed and “hidden” layer of training data representations in the bottleneck between the algorithm’s encoder and decoder). Assessing the history of machine learning alongside these emerging online social ecologies we will home in on a problem first acknowledged as early as 1976 yet largely unexplored in the literature: following from bio-inspired frameworks these neural network architectures unwittingly replicate a theory of the imagination already described by Kant in CPR. By examining the reception of Kant’s transcendental imagination in contemporary philosophy, it will be argued that cognition’s encounters with AI are non-trivial for understanding the modelling procedures of the imagination in art, science, or politics as consequential anticipatory systems.
Keith Tilford is an artist and writer living in Brooklyn, NY. He is illustrator of the graphic novel Chronosis, co-created with Reza Negarestani and Robin Mackay (Urbanomic, 2021), and recent writings have appeared in Glass Bead Journal and Superpositions: Laruelle and the Humanities (Rowman & Littlefield, 2017).
Reading: We will be reading Pietro Perconti and Alessio Plebe, "Deep Learning and Cognitive Science".