This study compares how humans and LLMs handle focus particles like 'even' and 'only' using a 100-item dataset of scalar judgments. Preliminary findings suggest that while outputs may appear similar, the underlying cognitive and computational mechanisms differ.
HOW THIS AFFECTS YOU
●
researcherThis highlights potential risks in assuming LLM semantic reasoning mirrors human linguistic logic.