Tianyu Jiang

Last-utterance clarifications can worsen parsing accuracy, with parser-agnostic rewrites introducing more errors than fixes in real-world applications.

LLMs' moral compasses are surprisingly swayed by their feelings: inject a little joy and suddenly previously unacceptable actions get a pass, revealing a critical divergence from human moral reasoning.

LLMs can distinguish between literal and figurative meanings early in their processing, revealing a surprising geometric structure that simplifies figurative-language classification.

LLMs can detect rhetorical questions with surprisingly high accuracy, but that doesn't mean they understand rhetoric the way we do.