Multi-Agentic AI for Fairness-Aware and Accelerated Multi-modal Large Model Inference in Real-world Mobile Edge Networks

Researchers developed a system using multiple AI agents to efficiently run large AI models on phones and local edge networks, not centralized servers. It tackles slow speeds and uneven performance by intelligently routing tasks and managing resources. In tests, it cut average delays by over 80% and significantly improved fairness of service. The system adapts quickly without needing retraining, offering a scalable solution for faster, more private AI on mobile devices.