LLMs reliably reproduce the qualitative structure of human social inferences on numerical precision but vary in magnitude calibration, with combined pragmatic prompting improving all metrics across models.
We assess alignment at three levels
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Social Meaning in Large Language Models: Structure, Magnitude, and Pragmatic Prompting
LLMs reliably reproduce the qualitative structure of human social inferences on numerical precision but vary in magnitude calibration, with combined pragmatic prompting improving all metrics across models.