When Too Many Cooks Spoil the Broth: Three Failure Modes of Multi-Agent LLM Reliability
A three-phase framework treating self-confidence assessment, peer-resistance capability, and error correlation as orthogonal axes requiring distinct training signals is proposed, and query-inherent rarity, not shared training, is identified as the dominant driver of correlated failure.
Aradhana Rai, V. K. Madisetti
· ACM AI Letters · 0 citations