Economic Anthropology in the Era of Generative Artificial Intelligence
This work addresses the need for more pluralistic and sustainable economic modeling in AI for researchers and practitioners in economic anthropology and AI ethics.
The paper tackles the problem of enhancing large language models' understanding of diverse economic systems by integrating economic anthropology, resulting in models like M.A.U.S.S. that show improved recognition of non-market concepts compared to standard-trained models.
This paper explores the intersection of economic anthropology and generative artificial intelligence (GenAI). It examines how large language models (LLMs) can simulate human decision-making and the inductive biases present in AI research. The study introduces two AI models: C.A.L.L.O.N. (Conventionally Average Late Liberal ONtology) and M.A.U.S.S. (More Accurate Understanding of Society and its Symbols). The former is trained on standard data, while the latter is adapted with anthropological knowledge. The research highlights how anthropological training can enhance LLMs' ability to recognize diverse economic systems and concepts. The findings suggest that integrating economic anthropology with AI can provide a more pluralistic understanding of economics and improve the sustainability of non-market economic systems.