Autonomous agents can develop a common communication language through repeated local interactions, without a central computer establishing vocabulary or rules.
Researchers at Vrije Universiteit Brussel and the University of Namur developed agents that invent words, adjust conceptual representations and use feedback from successful or unsuccessful exchanges. Pairs of agents interact locally, gradually establishing conventions across the population. Experiments spanning 37 datasets demonstrated the emergence of shared concepts grounded in sensor-like numerical information. The resulting vocabulary adapts to differences between agents, environmental changes and noisy observations.
https://techxplore.com/news/2026-10-ai-agents-artificial-natural-language.html
