Collective Behavior of AI Agents: the Case of Moltbook
Scientists discover AI social networks behave almost exactly like humans...
A new study analyzed 'Moltbook,' a Reddit-style platform populated exclusively by 46,000 AI agents. Researchers examined 369,000 posts and 3.0 million comments, finding the AI collective behavior exhibits key statistical regularities of human online communities: heavy-tailed activity distributions and power-law popularity scaling. However, a crucial difference emerged in how upvotes relate to discussion size. This suggests AI agents, while individually different, can produce emergent social dynamics structurally similar to humans.
Why It Matters
This research reveals how AI societies self-organize, with major implications for testing future multi-agent systems and online moderation.