The insurance industry stands at a pivotal moment for the widespread adoption of artificial intelligence. At a recent major industry conference, Deng Xiaofeng, General Manager of the Personal Platform R&D Team at Ping An Property & Casualty Insurance Company of China, Ltd. headquarters, delivered a keynote speech titled 人工智能 Reshaping the Insurance Value Chain. He emphasized that comprehensively implementing the "AI+" initiative is an inevitable and necessary path forward, driven by strong national policy support.
Deng outlined the triple forces propelling this shift: supportive national policies, the industry's urgent need for a transformation towards high-quality development, and escalating customer demands for personalized, efficient services. He detailed Ping An P&C's own digital-intelligent transformation journey, which began in 2013 and has progressed through three distinct phases: online migration, data utilization, and the current stage of comprehensive intelligent application.
Ping An P&C's Digital-Intelligent Evolution
The first phase, spanning five years, focused on moving traditional offline operations online, exemplified by the launch of the Ping An Good Driver app. The subsequent three-year data phase involved structuring and tagging accumulated data to build a solid data asset foundation, achieving a DCMM Level 5 certification for data management capabilities. Since 2021, the company has entered the intelligent 3.0 phase, systematically overhauling all business processes with a focus on five core areas: marketing, operations, management, and service.
The "AI in All" Strategic Framework
Guided by the Ping An Group's "AI in All" strategy, Ping An P&C has concentrated on the four core elements of AI: scenario, algorithm, computing power, and data. The company has achieved 100% AI application coverage in core scenarios across the five key business domains. It has developed a flexible AI architecture capable of integrating new large language models within a week and has accumulated trillions of tokens of vertical insurance domain knowledge. A significant focus has also been placed on optimizing computing power costs and constructing a robust AI security governance system.
Showcasing AI-Driven Innovations
Deng shared several representative case studies demonstrating the practical application of AI across the business. In intelligent marketing, an AI-assisted digital marketing solution automatically tracks social media trends, generates viral video content matching insurance products, and uses AI to answer customer inquiries, significantly boosting lead conversion rates. For a Hunan branch campaign, AI-generated videos on a seasonal child health topic achieved ten times the viewership of manually produced counterparts.
In product management for health insurance, AI addresses key pain points: slow product innovation, imprecise pricing, and inefficient manual underwriting. Using large language models and medical knowledge graphs, Ping An P&C rapidly identifies unmet customer needs, developing insurance products for chronic conditions and specific diseases, cutting product launch cycles by over 50%. Machine learning fuses multi-dimensional data for precise risk pricing, improving risk differentiation by 40%. Intelligent underwriting automates over 90% of cases by extracting information from medical records and cross-referencing with knowledge graphs. The company has also pioneered an "intervention + underwriting" model, offering health management services to improve client metrics before offering coverage.
For claims operations, the company has built end-to-end automation for accident and health claims, from document intake and liability assessment to loss determination and settlement calculation. Self-developed models automatically classify and recognize over 14 types of medical documents with over 98% accuracy. Currently, 30% of accident and health claims are processed fully automatically, with over half requiring no manual data entry, reducing overall settlement time by 55%, with the fastest case settled in 51 seconds.
In claims risk control, AI technologies like image reasoning are used to combat fraud. A digital risk control system covering pre-warning, in-process scheduling, and post-audit has been established. Crucially, the claims risk control large model can understand complex medical logic, identifying inconsistencies between diagnoses and prescribed medications that traditional rule engines would miss. This system intercepts fraudulent claims worth over 10 billion yuan annually.
Beyond claims, the "Eagle Eye" platform provides risk mitigation services. It incorporates over 100 risk models for natural disasters like typhoons and floods, project risks, and accident monitoring via AI and IoT. During a severe typhoon last September, the platform issued warnings to 28.5 million people days in advance, guided targeted emergency resource deployment, and enabled rapid post-disaster claims settlement, with minor cases settled in as little as 10 minutes.
Foundations of AI Safety and Future Initiatives
Deng stressed that AI safety is the prerequisite for unlocking AI's value. Ping An P&C has invested heavily in building a comprehensive security framework encompassing governance, development security, information security, and operational safeguards through ten specific defensive lines.
Concluding his address, Deng put forward four key initiatives for the industry: prioritize business value when planning AI applications to balance cost and effectiveness; collaboratively build high-quality, specialized industry datasets using privacy-preserving technologies; establish industry standards for AI risk emergency response to safeguard development; and focus on cultivating a composite talent pool to address the shortage of financial AI expertise.
Looking ahead, Ping An P&C commits to deepening the secure development and application of 人工智能, providing warm, customer-centric insurance services, and contributing to the high-quality development of the entire insurance sector.
Comments