Decoupling Human-AI Collaboration for the Application of Large Models in the Power Industry

As large models continue to gain influence, their application is expanding beyond everyday life and work into traditional industries, including the power sector. The power industry, with its high demands for precision in mechanisms and logical reasoning, presents challenges for large models, which are still evolving and prone to issues like hallucination—weaknesses that cannot be tolerated in critical power systems. To facilitate the practical deployment of large models in the power industry, it is essential to decouple the stable components of artificial intelligence (AI) technologies, such as application scenarios, knowledge bases, and intelligent agents, from their rapidly advancing elements, like models and computational power. One promising approach to achieve this decoupling is through the establishment of Human-AI Collaborative Business (Double Intelligence Collaboration). In this framework, tasks that are well-suited for AI are handled by AI with human oversight, while tasks that AI is not yet capable of autonomously managing are performed by humans with AI assistance. This paper explores how such a decoupling strategy, through human-AI collaboration, can ensure both high reliability and effective application of AI technologies in the power industry, facilitating their integration while minimizing risks. At the same time, it leaves room for future opportunities to adopt rapidly evolving AI technologies.