Generative artificial intelligence-oriented synthetic network: Toward Integrated Fine-Tuning and Inference When Generative Artificial Intelligence Meets Edge Intelligence in the Intelligent Internet of Vehicles

Generative artificial intelligence (GAI) and edge intelligence (EI) are driving the evolution of traditional vehicular networks toward the intelligent Internet of Vehicles (IIoV) by providing foundational and personalized knowledge. However, their inherently contradictory characteristics present significant challenges to direct knowledge sharing. Furthermore, conventional methods that independently optimize fine-tuning and inference lack the foresight to achieve long-term network benefits. To address these challenges, we propose the GAI-oriented synthetic network (GaisNet), a collaborative cloud-edge-end intelligence framework that integrates fine-tuning and inference. GaisNet, specifically, can mitigate contradictions by leveraging data-free knowledge relays, where bidirectional knowledge flow facilitates a virtuous cycle of model fine-tuning and task inference with a long-term perspective, fostering mutualism between GAI and EI in the IIoV. A case study illustrates the effectiveness of the proposed mechanisms. Finally, we discuss the future challenges and directions in the interplay between GAI and EI.