Unmasking conversational bias in AI multiagent systems
Abstract
Detecting biases in the outputs produced by generative models is essential to reduce the potential risks associated with their application in critical settings. However, the majority of exi…