Agentic AI as Co-Comparatist: Autonomous Literary Intelligence and The Posthuman Turn in World Literature Studies

This conceptual paper asks a narrow question with wide consequences. When an artificial intelligence system stops behaving like a retrieval instrument and starts decomposing goals, selecting its own tools, and issuing interpretive statements without a prompt for each move, does anything in comparative literature change, and if so what, and under what conditions? The discipline still tends to treat AI as a search or translation aid, while the technical literature has moved to systems that act across several steps with little supervision. The paper does not argue that machines read, understand, or enjoy literature. It defends a deliberately limited proposition: that such a system can take on the functional role of a co-comparatist, proposing intertextual relations across corpora that human scholars must then evaluate, and that this functional shift is enough to unsettle the anthropocentric foundation of comparative method. The method is theory adaptation applied to a structured corpus of ninety recent sources on agentic AI, set beside foundational scholarship in world literature, translation studies, and digital humanities that carries no date restriction, and read through posthumanist theory and agential realism. The paper separates three things: machine pattern detection, machine-assisted interpretation, and autonomous scholarly interpretation. It offers a claim-level evidential audit, a four-layer governance cycle in which marginalised communities participate from corpus design onward, a worked example tested under favourable and adverse conditions, and eight conceptual propositions with proposed measures. The reverse risk is weighed equally: a corpus skewed toward Western canons would automate Eurocentric hierarchy at scale.