A new framework derives testable consistency conditions between elicited probabilities and decisions without assuming any utility function, revealing small but detectable mismatches in LLMs on medical tasks.
URL http://www.jstor.org/stable/1914185
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
verdicts
UNVERDICTED 2representative citing papers
People reject cookie + $2 offers from robots more than cookie alone due to inferred phantom costs, accepting more from robots than humans overall with no embodiment effect for robots.
citing papers explorer
-
When Agents Say One Thing and Do Another: Validating Elicited Beliefs from LLMs
A new framework derives testable consistency conditions between elicited probabilities and decisions without assuming any utility function, revealing small but detectable mismatches in LLMs on medical tasks.