He Dongchuan, General Manager of the Information Technology Department at PICC Life Insurance, recently stated that when discussing AI algorithm recommendations, the industry must first clarify the business landscape of internet insurance. He made these remarks during a closed-door seminar on "AI Insurance Marketing: Compliance Boundaries and Institutional Transformation" hosted by the National Business Daily.
According to He, the primary traffic for internet insurance is currently concentrated on large platforms, with insurance companies' own direct sales channels accounting for a relatively limited share. Even in direct sales operations, the application scenarios for algorithm recommendations are scarce, as the customer base mainly comes from major internet platforms. Consequently, when collaborating with these platforms, insurers find themselves in a weak position for information acquisition, with very limited data being transferred to them. This is the fundamental reality of internet marketing in the insurance industry today.
"As for direct sales channels, insurance institutions will inevitably make differentiated recommendations to customers. The core logic lies in suitability management—matching products that align with a customer's risk level based on their consumption capacity and history, which is similar to the risk-matching principle in stock investments," He explained. He added that algorithms optimize the efficiency of the sales funnel, reflecting professional service. Regarding the market's focus on large model customer-facing applications, they are still in the pilot phase.
"Currently, only four financial institutions have been approved to conduct pilot programs for large model customer-facing applications, and these pilots are limited to Suzhou and Guangzhou," He noted. He said that customer-facing large models build a private knowledge base based on an industry knowledge foundation, leveraging the underlying large model for intelligent interaction. The pilot scope may expand in the future, but the boundaries must be strictly observed at this stage.
From a technical perspective, He further elaborated that large models can replace manual work in batch reviews for compliance and audit processes. Companies can build a dedicated knowledge base from regulatory rules and internal policies to automatically verify marketing copy and client materials. However, the biggest obstacle to implementation is the allocation of responsibility, as AI cannot bear the consequences of compliance errors. If any oversight occurs in the review, the insurance institution will be held liable. Therefore, in human-machine collaboration scenarios like AI-generated marketing copy, all output must be reviewed by the institution and confirmed by licensed marketing personnel before use, clarifying the institution's primary management responsibility and avoiding false advertising risks.
"I believe the core value of large models is to amplify existing capabilities, not to create something out of nothing. Companies must first build a solid knowledge base, including business rules and best practices, which is the foundation for applying AI, rather than blindly pursuing the model itself," He stated. He explained that the large model base will become highly centralized in the future. Insurance companies will select mature foundations and inject their own proprietary knowledge to empower internal management and marketing. "What we can do at this stage is to focus on building knowledge bases for products, compliance, and customer service. By using AI to simplify repetitive tasks like plan creation, product comparison, and content generation, we aim to transform agents into professional super-advisors, rather than pursuing a fully unmanned sales process."
"From a long-term industry development perspective, as regulators continue to tighten the rules on individual independent online sales, online AI marketing must be centrally coordinated, reviewed, and fully recorded by the institution. AI will accelerate the industry's institutional transformation, but it will not eliminate individual agents. The model of solo operations will persist for a long time, and AI will only significantly enhance the service efficiency of individual agents," He said. He emphasized that the essence of life insurance is trust and warmth. Unlike standardized car insurance, life insurance sales rely on the emotional connection between people. AI can help agents improve efficiency and enhance professionalism, but it cannot replace the face-to-face trust-building by individual agents. Even with the advent of general artificial intelligence, the uniquely human abilities of perception, emotion, and empathy will remain irreplaceable. Therefore, the industry should focus on leveraging AI to amplify the capabilities of agents, rather than fantasizing about completely replacing human sales.
Comments