OneConnect International CTO at Fortune Brainstorm AI: Future of Finance Demands Hybrid AI Approach for Trust and Control
SINGAPORE, July 28, 2025 /PRNewswire/ -- The future of artificial intelligence in finance lies in a hybrid approach that combines the generative power of large models with the precision of traditional AI to ensure trust and control, said Dr. Ming Jin, International CTO of OneConnect Financial Technology (NYSE: OCFT; HKEX: 6638), at the Fortune Brainstorm AI conference in Singapore. Dr. Jin explained that this integrated strategy is crucial for sectors like finance, which handle sensitive data and require high levels of reliability.
Dr. Jin pointed out that the banking industry commonly faces challenges such as complex legacy systems and aging architectures. Artificial intelligence, especially large language models, offers new solutions for modernizing systems. "By using AI large models to analyze program code and extract functional specifications, we can quickly understand and reconstruct the underlying architecture," he said. This AI-enabled approach is proving to be an effective tool for financial institutions to overcome system modernization challenges.
He cited the success of OneConnect's parent company, Ping An Group (HKEX: 2318; SSE: 601318), as an example of AI in practice. The Group has utilized AI in multiple core scenarios including sales, service, operations, and management. In 2024, Ping An's AI call center handled 1.84 billion customer interactions, accounting for 80% of total inquiries. In risk management, large model-powered anti-fraud technologies have achieved significant results in key areas such as credit and fraud prevention. Notably, in identifying forged facial information and tampered identity data, these algorithms provide real-time analysis, building a robust security defense and substantially reducing operational risks and costs.
Discussing the global landscape, Dr. Jin noted key differences in AI deployment. Financial institutions in mainland China generally adopt private deployments and focus on developing financial domain-specific models based on open-source frameworks, while those in Hong Kong SAR and Southeast Asia often build training and iteration systems based on commercial general large models within regulated public cloud environments. These differences affect model training, data governance, and service delivery in concrete ways.
Dr. Jin emphasized three key elements for successful AI implementation: first, fundamental capabilities based on data, computing power, and algorithms; second, professional talent with cross-disciplinary expertise in finance and technology; and third, a rich and continuously evolving portfolio of business scenarios. He highlighted China's significant advantages in these areas, especially in accumulating data-intensive scenarios and a strong supply of interdisciplinary talent, providing a solid foundation for the broad application of AI technology in finance. For example, Ping An has more than 3,000 scientists and over 21,000 technology developers, and through over a decade of continuous investment in AI, has realized integrated applications of traditional AI and large model technologies. To address reliability and trust issues in large model content generation, Ping An has pioneered a framework combining "large models + precise inference from traditional AI + human decision nodes" to ensure transparency and interpretability in technology use.
Regarding the future direction of AI development, Dr. Jin stated that although large models show great potential in content generation, industries like finance--which require higher trust and controllability--still need to combine them with traditional AI. OneConnect is exploring the integration of generative models into business processes while retaining key points for human intervention to ensure system flexibility, transparency, and reliability.
"We always believe AI is a tool whose value lies in scalability and scenario adaptability," Dr. Jin said. Currently, Ping An Group has successfully built a three-tier large model system comprising general models, vertical domain models, and application models. It has established five major laboratories and nine databases, making it one of the world's largest financial and healthcare data repositories. AI products and solutions are widely applied across 85 large model scenarios within the group, accelerating the development of its ecosystem.
Looking ahead, OneConnect, as Ping An's sole platform for exporting financial technology, is committed to advancing its product development around large models and related AI capabilities in sales, service, operations, and management. It aims to promote the application of AI technology in financial services and, leveraging its "platform + ecosystem" model for international expansion, systematically export platforms and scenarios in overseas markets, contributing Ping An's technology expertise to the digital transformation of global financial institutions.