Multi-agent discussion gains less when dissent is withheld
Multi-agent systems of LLMs add discussion to majority voting and are therefore expected to be more capable. However, empirical reports conflict on whether discussion improves accuracy or leads to an incorrect consensus. Here, we introduce a parsimonious model that explains when discussion improves accuracy and when it...
Chand Sahil Mansuri, Xin Wang, Meng-Ying Li et al.
· 0 citations