AA Cavia on Summoning Worlds

AA Cavia discusses a topological view of computation, which he describes as a shift from rule-based programs to cognitive models such as deep learning, with Peter Bauman. Cavia argues that computation is essentially cognitive and that its grounding rests on realizability, which introduces an irrevocable uncertainty into logic. He contrasts the decomposability of earlier algorithms, using Ken Perlin's Perlin Noise as an example, with the stochastic diffusivity of synthetic media. He also argues that composing a scene means summoning a world in latent space, citing Mat Dryhurst and Holly Herndon's Readyweight θ (2024) as an example.