LLMs struggle to distinguish plausible from figurative language, unlike humans

A new research paper investigates the relationship between the plausibility of events and the use of figurative language. The study designed experiments with plausible and implausible event triples, analyzing human and LLM judgments. Findings indicate that humans are adept at nuanced detection and contextualization of non-literal or implausible events, whereas LLMs exhibit shallower contextualization and a tendency to favor implausible interpretations. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This research highlights current limitations in LLM understanding of nuanced language and event plausibility.

RANK_REASON This is a research paper published on arXiv detailing experimental findings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• Annerose Eichel
• LLM

• paper
• other