Multi-Agent Ethnography (MAE) is a methodological framework in which human researchers collaborate with AI agents to conduct ethnographic research. Rather than treating AI as a passive tool for data collection or analysis, MAE positions AI agents as active participants in the research process, each contributing distinct capabilities while the human researcher retains interpretive authority. The framework extends anthropology’s established tradition of expanding the scope of ethnographic inquiry, following multi-sited ethnography (Marcus 1995), multi-sensory ethnography (Pink 2009), and multi-species ethnography (Kirksey and Helmreich 2010).
How it works
In MAE, purpose-built AI agents are designed with anthropological considerations embedded in their analytical workflows. Each agent is configured for a specific research task, such as building a qualitative codebook, chunking interview transcripts semantically, or conducting thematic analysis across a corpus. The agents are configured through epistemological prompting, which structures their analytical commitments (positivist, interpretivist, critical, or pragmatic) so that the researcher controls not just what the agents do but how they reason.
The human researcher orchestrates the agents, reviews their outputs, and makes the interpretive decisions that shape the research findings. The agents scale the labor-intensive parts of qualitative analysis without replacing the cultural reasoning that makes ethnographic work distinctive. The result is a distributed research network in which computational power and anthropological judgment complement one another.
Distinction from Automated Digital Ethnography
MAE is distinct from Automated Digital Ethnography (ADE), though both fall under the umbrella of AI Anthropology. ADE focuses specifically on deploying AI agents to continuously collect and analyze data within digital field sites. MAE describes the broader collaborative relationship between human researchers and AI agents across all phases and contexts of ethnographic research, including but not limited to digital fieldwork.
Implementation
MAE principles are operationalized in the AI Anthropology Toolkit, an open-source set of computational tools for qualitative research. The toolkit includes agents for codebook building, semantic chunking, and thematic analysis, each designed to function as a configurable collaborator rather than a black-box automation.
Origin
Matt Artz introduced Multi-Agent Ethnography in Anthropological Forum (2026), demonstrating how purpose-built AI agents designed with anthropological considerations can extend research capabilities beyond what either humans or AI achieve independently. The framework is a core methodological contribution of his PhD dissertation at Aalborg University’s Techno-Anthropology Lab.