Consensus online is hard...for humans

I prefer to call them "bots" -- computer code that has the capacity to take actions online -- rather than "agents." We humans are already biased toward anthropomorphising -- especially in regard to entities that can emulate human verbal communication (see The Media Equation). We are seeing emerging behaviors like collusion and consensus -- I don't know if a psychology of LLMs is possible, considering the fact that humans are not yet able to determine exactly how models reach particular conclusions, but it looks like a sociology of bot swarms should be a viable area of study

AI agents can build consensus on a scale humans can't

Everyone is familiar with the situation: A larger group of people plans to visit a restaurant together, but it can take time and sometimes a great deal of patience to agree on a time and place to meet. The better the participants know each other and their preferences, the faster they will reach an agreement. However, past social science experiments have demonstrated that people are only able to socialize effectively with 150–200 others without requiring rigid rules. Researchers call this Dunbar's number.

A team of researchers from the University of Konstanz has now documented that artificial intelligence (AI), too, has a Dunbar number—although it is significantly larger. When AI agents of large language models (LLMs) are asked to reach a consensus, this is possible with up to 1,000 individual agents—depending on the model. The findings are published in the journal Science Advances.