Towards Edge Intelligence

Title

Why AI Still Can't Handle the Physical World: AGI-Native Wireless Systems for Physical AI

Abstract

Artificial intelligence has transformed many sectors, yet today's AI tools, including deep learning and generative AI, still struggle when embedded into physical systems such as robots, drones and vehicles operating under real-world physical laws. This talk addresses three intertwined challenges: limited ability to handle unseen and out-of-domain scenarios, lack of first-principle solutions for physical agents navigating the real world, and the need for pervasive connectivity to support physical AI tasks at scale.

The talk introduces artificial general intelligence (AGI)-native wireless systems, a wireless architecture and framework for supporting the intelligence and communications needs of physical AI systems. It explores a fusion of wireless systems, digital twins, neuroscience and AI through an AGI architecture based on perception, world model and action-planning, with cognitive capabilities such as reasoning, planning, imagination and deep thinking. The talk discusses semantic communication for physical AI, a world model architecture inspired by thinking fast and slow, and test-time scaling laws that help physical AI agents handle unforeseen real-world scenarios.

Bio

Walid Saad received his PhD degree from the University of Oslo in 2010. He is currently the Rolls Royce Commonwealth Professor in Digital Twin Technology, a Professor in the Department of Electrical and Computer Engineering, and a founding faculty member of the Institute for Advanced Computing at Virginia Tech, where he leads the Network intelligEnce, Wireless, and Security laboratory. His research interests include wireless networks, machine learning, game theory, quantum communications and learning, security, UAVs, semantic communications, cyber-physical systems and network science. Dr. Saad is a Fellow of the IEEE and Editor-in-Chief of the IEEE Transactions on Machine Learning in Communications and Networking.