When anthropologist Robin Dunbar introduced the idea of Dunbar’s Number in the 1990s, he revealed a fascinating limit to our human social capacity: the average person can maintain stable relationships with around 150 people. This cognitive limit shapes how we build our social circles, teams, and organizations. Beyond 150 connections, things start to break down—communication falters, collaboration weakens, and it becomes difficult to maintain personal connections.
But AI? It turns out, AI models don’t have this problem.

Recent research has shown that artificial intelligence models—like those behind large language models (LLMs)—can collaborate and reach consensus in groups far larger than we humans could ever manage. We’re talking about thousands of AI models working together seamlessly to solve problems, coordinate projects, or process vast amounts of data.
How Do AIs Do It?
The study found that high-powered AI models like GPT-4 Turbo can cooperate with other AI agents at scale, functioning in massive “societies” that go beyond human limits. Just like how humans interact to reach consensus on decisions, these AI models can exchange information and adjust their decisions based on the group’s inputs. And unlike us, AI doesn’t forget details or lose track of conversations, because its memory is tied to its hardware, not cognitive limitations.
In one experiment, researchers ran multiple copies of AI models and assigned them tasks where they needed to reach consensus on questions with no obvious answer, like which side of the road a newly-formed country should drive on. Powerful AI models consistently came to an agreement, even when there were thousands of agents involved. For comparison, smaller AI models—those with less computational power—struggled to coordinate in much smaller groups.
What Does This Mean for Business?
The implications are huge, especially for industries relying on integration and automation. Traditionally, we’ve been limited by human capacity when it comes to organizing large teams, managing data, and making decisions across departments or entire organizations. AI, however, can work at scales that humans can’t even fathom, coordinating vast amounts of information and reaching conclusions faster and more accurately.
Imagine AI-driven systems that integrate and automate processes across an entire business, handling tasks that would normally take large teams to execute. With AI models working together at such scale, organizations could break through traditional limitations on problem-solving and collaboration, making it possible to solve highly complex challenges.
Challenges and Considerations
It’s not all straightforward, though. While AI models can coordinate at scale, they lack diversity in thought. Since many AI models are copies of the same system, they don’t bring different perspectives to the table like humans do. This could mean that while AIs can quickly reach consensus, they may miss out on creative or innovative solutions that come from diverse human teams.
Another challenge is ensuring that AI models reach decisions aligned with human values. AI systems operate based on data and algorithms, and they may not always reflect the ethical or social considerations we would expect from human decision-makers.
The Future of Collaboration with AI
Despite these challenges, the potential for AI in large-scale collaboration is undeniable. As AI technology continues to evolve, it will become an increasingly critical part of how we approach integration and automation. By leveraging AI’s ability to process and collaborate at scale, businesses can unlock new efficiencies, handle larger projects, and solve more complex problems.
AI offers the chance to extend what’s possible, allowing businesses to operate smarter, faster, and at a scale we’ve never seen before.
Ready to Explore What’s Possible?
If you’re curious about how AI-driven collaboration can transform your business processes and streamline your automation efforts, reach out to us. We’d love to discuss how AI can expand your potential and help you achieve more.
Further reading