Stop Making Sense: Complexity and Contradiction in AI and Architecture

Summary

Within generative AI, programmers are driven to make models as coherent as possible, as coherence is a crucial characteristic of a productive and profitable AI model. AI companies often improve their models by employing aesthetic models that assign image weights based on perceived aesthetics. This often involves human evaluators scoring images to train the model. AI models are inherently sketchy; they are the model's predictions of what the user wants to see. Nonsense, ambiguity and incoherence have long been traditions within literature, art and music. Nonsense allows for exploring uncharted territories, challenges conventions, and generates novel ideas. As we navigate the intersection of AI and creativity, it is crucial to strike a balance between coherence and creative nonsense to harness the potential of AI as a tool for exploration, dreaming and the creation of new architecture.