Interoperable AI for Self-Organisation

Knowledge or information sharing is one of the foundational principles shared by the Web and any well functioning decentralised multi-agent system (MAS). Another is interop-erability. Engineers of MAS follow design strategies for ensuring interoperability that range from using a common ontology to using translation mechanisms. But what kind of sustainable communication strategy would AI agents themselves use if given the chance to self-organise? In this short paper we summarise our previous findings from [6], where we used an evolutionary agent-based modelling approach to answer this question.