Send message to Author

Paper short abstract

Drawing on two years of building AI tools for anthropological research, this paper examines a paradox: when anthropological knowledge is successfully infrastructured into computational systems, it becomes invisible. If expertise is no longer legible as labor, what becomes of the knowledge worker?

Paper long abstract

Debates about AI and the future of work often frame the question as replacement or resistance. This presentation argues for a third position: that large language models enable anthropologists to execute creative and analytical work previously constrained by time, technical skill, or resource limitations. This latent creativity becomes actionable through multi-agent ethnography (Artz 2026), an approach that positions AI agents as configurable collaborators within distributed human-AI research networks.

MAE extends anthropology's tradition of methodological innovation, from multi-sited to multi-species to multi-scalar ethnography, by incorporating AI agents as research partners coordinated through conversational interaction. Drawing on two years of developing purpose-built AI tools for anthropological research, I demonstrate how disciplinary reasoning can be partially encoded into computational infrastructure while preserving human interpretive authority.

Yet this raises a critical question for post-work futures: if anthropological knowledge can be infrastructured (Scott 1998; Hughes 1989, 1983) into AI systems, what remains as distinctly human labor? The answer is uncertain, complicated by a paradox: when expertise is successfully encoded, it becomes invisible (Bowker et al. 2010; Star and Ruhleder 1994), potentially undermining the conditions that make expertise legible as labor. If this happens, the question becomes not what work remains, but what becomes of the worker when their expertise no longer requires them.