Towards Foundation Models with Native Multi-Agent Intelligence
This work addresses the challenge of enabling foundation models to effectively operate in multi-agent contexts, which is crucial for advancing AI systems in collaborative and interactive environments, though it is incremental as it builds on existing single-agent capabilities.
The paper tackles the problem that foundation models lack robust multi-agent intelligence despite strong single-agent performance, providing empirical evidence across 41 large language models and outlining research directions to address this gap.
Foundation models (FMs) are increasingly assuming the role of the "brain" of AI agents. While recent efforts have begun to equip FMs with native single-agent abilities -- such as GUI interaction or integrated tool use -- we argue that the next frontier is endowing FMs with native multi-agent intelligence. We identify four core capabilities of FMs in multi-agent contexts: understanding, planning, efficient communication, and adaptation. Contrary to assumptions about the spontaneous emergence of such abilities, we provide extensive empirical evidence across 41 large language models showing that strong single-agent performance alone does not automatically yield robust multi-agent intelligence. To address this gap, we outline key research directions -- spanning dataset construction, evaluation, training paradigms, and safety considerations -- for building FMs with native multi-agent intelligence.