Emergent language among AI agents: a path toward energy efficiency and water conservation

The rapid expansion of artificial intelligence (AI) poses an increasing dilemma: its enormous energy and water consumption threatens environmental sustainability in a global context of climate crisis and resource scarcity. Although the use of more efficient hardware, renewable energy, and optimization techniques has improved data center efficiency, these solutions primarily focus on infrastructure, leaving unexplored the potential of communication among AI agents themselves. Recent developments in emergent languages show that agents can create autonomous protocols to coordinate more efficiently, reducing computational redundancies and data transmission, with direct implications for lowering the energy and water required for cooling and processing. In this perspective review we argue that optimized communication between agents represents a complementary pathway to align technological advancement with sustainability, while maintaining or even improving system performance. Advancing this field requires designing standardized metrics that integrate performance and environmental footprint, testing these protocols in real-world resource-limited scenarios, and establishing regulatory frameworks that make their impact transparent. This approach could transform the relationship between AI and sustainability, guiding it toward a more resilient and responsible future.