Bricolage, Wild Thought, and the Automation of Knowledge

In Wild Thought (1962), the French anthropologist Claude Lévi-Strauss challenges Eurocentric perceptions of scientific epistemological frameworks as uniquely valid. He points to the existence of what he calls “wild thought”: a mode of cognition that is semi-autonomous and which self-seeds, putting out roots wherever it falls. Among the material manifestations of wild thought is bricolage: a term which appears frequently in histories of computing to describe the practices of hackers and innovators. This chapter aims to recover Lévi-Strauss’s lexicon of wild thought and bricolage for artificial intelligence. Traversing anthropology, contemporary media theory, and new experiments in online subcultures, I argue that the epistemology at work in Large Language Models (LLMs) bears a striking resemblance to frameworks intuited by Lévi-Strauss. Ultimately, I suggest that Lévi-Strauss’s concepts of bricolage and wild thought might offer fresh lenses through which to explore the cognitive ecologies of LLMs.