Gao Feng, Chairman of Shenzhou Guangda Technology, said at the Fifth China Original Economics Forum and Economists & Entrepreneurs Summit held on September 6 that the essence of AI is a high-level compression of existing knowledge. He cautioned that over-reliance on general-purpose large models could strip companies of their autonomous iteration capabilities and widen the competitive gap.

General-purpose models are trained on publicly available internet data, which is mixed and noisy. As a result, they cannot adapt to specialized scenarios in niche industries or solve real operational challenges for businesses. Gao Feng noted that many companies currently equate going online with becoming intelligent, resulting in fragmented systems and superficial applications that fail to truly activate the value of digitalization.

He stressed that the core of new quality productive forces lies in the dual improvement of efficiency and benefits brought by technology implementation. The critical path for enterprise AI transformation, he said, is to create an exclusive, proprietary intelligent system. The experience, expertise, methodologies, and industry Know-how accumulated over years of deep engagement in a sector are irreplaceable core assets.

Enterprises should build autonomous post-learning intelligent systems that convert their own industry experience into proprietary training corpora, develop private AI models, and rely on intelligent agents to achieve autonomous decision-making, automated office operations, production optimization, and marketing efficiency. This approach comprehensively reduces labor and operational costs while enhancing the added value of products and services.

Gao Feng concluded that amid intensifying K-shaped divergence, those who can establish a dedicated enterprise AI brain and solidify industry data moats will escape homogeneous competition and secure an upward trajectory in technological iteration and industrial transformation.