Multi-agent teams might not be better than a single good model: Apple and Stanford paper

A recent study by Apple and Stanford reveals that in certain scenarios, multiple agents may not outperform a high-quality single model, challenging the assumption that more is always better in AI and machine learning configurations.