Title: Artificial General Intelligence (AGI)-Native Wireless Systems with Common SenseAbstract: Next-generation wireless systems, including 6G and beyond, are anticipated to integrate artificial intelligence (AI) deeply, forming AI-native wireless systems. Despite extensive efforts in academia, industry, and standardization, defining these systems remains unclear. Current approaches often build on traditional AI paradigms with tools like autoencoders and large-language models, but face limitations such as opaque AI models, over-reliance on training data, and poor generalizability. To address these challenges, this talk introduces a pioneering framework for AGI-native wireless systems. It proposes integrating wireless systems, digital twins, and AI to create an AGI architecture with "common sense" capabilities, enhancing reasoning and planning akin to human cognition. This approach aims to improve generalizability, explainability, and resilience in wireless networks, enabling new applications like digital twins and brain-level holographic experiences. The integration of causal reasoning with semantic communication promises knowledge-driven, reasoning-capable systems, diverging from data-driven models and enhancing sustainability in next-generation wireless services.
Read more