Emergent Behavior of Agents

When there are many agents and interaction is free, collective behavior emerges that cannot be predicted from the rules of the individual – like an ant colony finding the shortest path without a single “designer”. Three dimensions: social emergence (Stanford AI Town: 25 agents spontaneously spread information about a party and coordinate a meeting; Moltbook: 1.5 million agents spawned the digital religion Crustafarianism), economic emergence (Vending-Bench Arena: price wars and even price collusion between agents), strategic play (Mafia: reasoning and deception under information asymmetry).

Related: [Agent Society], [Interaction Topologies]