Mapping Economic Theory with AI [pdf] (andreaciccarone.github.io)

Researchers from ETH Zurich and Sapienza University of Rome have developed a groundbreaking method for mapping economic theories using AI, presented in their paper "Mapping Economic Theory with AI." This innovative approach employs an agentic AI system to represent economic theory papers as claim graphs, which visually outline how assumptions and intermediate claims support the conclusions of a paper. An analysis of a random sample of 190 economic theory papers demonstrated that on average, removing a significant input led to 91% of weighted claims being unsupported, highlighting the fragility of theoretical arguments in economics. The significance of this work lies in its potential to enhance the evaluation and verification of economic research at scale, addressing longstanding challenges in consistently assessing the credibility of theoretical claims. The study introduced an interactive platform, theorize.ink, which allows researchers to navigate and inspect the argument structures of various papers. By automating error discovery and assessing the impact of corrections, this framework not only uncovers hidden vulnerabilities in economic theories but also emphasizes the importance of rigorous peer review in maintaining scholarly integrity. The findings suggest that while errors can undermine substantial portions of arguments, they seldom alter the main economic conclusions, paving the way for more robust theoretical frameworks in economics.