AI and Political Theory: Exploring Argument Mining for Normative Research—A Political Theorist's Perspective

This article examines how artificial intelligence—specifically, argument mining—can contribute to the methodological turn in contemporary political theory. While political theorists increasingly reflect on their methods, research designs, and forms of reasoning, discussions of argumentative structure remain largely implicit, rarely articulated within published work. At the same time, advances in AI, particularly large language models (LLMs), offer new tools for identifying argumentative patterns across large bodies of text. This study explores whether such tools can meaningfully support political theorists in clarifying, evaluating, and teaching normative argumentation. Focusing on argument mining, an NLP technique for detecting claims, premises, and inferential relations, the article presents an exploratory analysis of fifteen analytical political theory articles. Each article was coded twice—once by a human coder and once by an AI system—using a shared Toulmin-based framework consisting of Claims, Data, Warrants, Backings, Qualifiers, and Rebuttals. Precision, Recall, and F1-scores were calculated for each element and equally weighted across articles to allow systematic comparison. The findings highlight clear strengths, such as the consistent identification of claims and data, alongside persistent challenges, particularly in detecting context-sensitive elements like qualifiers. The analysis also underscores the need for human validation when dealing with dense, abstract normative texts. The article argues that AI-assisted argument mining cannot replace interpretive judgment, but it can illuminate recurring structures, support methodological transparency, and strengthen pedagogical clarity in political theory. By integrating computational tools with reflective normative inquiry, political theorists can expand their methodological repertoire while preserving the discipline’s commitment to argumentative rigor.