TLDR: The Financial Supervisory Service (FSS) in South Korea is set to deploy generative AI, akin to ChatGPT, to revolutionize its complaint handling. This initiative, part of a 33.4 billion won digital innovation project, aims to automate tasks like complaint classification, departmental recommendation, and drafting initial responses, significantly reducing the burden of over 100,000 annual complaints and accelerating processing times. The FSS acknowledges and plans to mitigate potential biases in the AI’s learning process.
The Financial Supervisory Service (FSS) of South Korea is embarking on a significant digital transformation, announcing plans to integrate advanced artificial intelligence (AI) technology into its complaint handling processes. This strategic move is part of the ‘Financial Supervision Digital Innovation Mid-to-Long-Term Project,’ a substantial 33.4 billion won initiative launched last month, aimed at creating an ‘AI secretary’ for its customer complaint handlers.
The core objective of this AI deployment is to streamline and expedite the resolution of the FSS’s substantial caseload, which currently exceeds 100,000 complaints annually. These complaints, ranging from unpaid insurance claims to loan rejections, often face processing times stretching up to six months. By delegating initial tasks to generative AI, similar to ChatGPT, the FSS anticipates a considerable reduction in staff workload and a significant acceleration in response times.
The AI’s workflow is designed to cover the entire complaint lifecycle. Upon receiving a complaint, the generative AI will first classify its type and recommend the most appropriate department for handling. Subsequently, it will generate a summary of the complaint by drawing upon similar past cases and relevant court precedents. Crucially, the AI will also be responsible for drafting initial response letters, which FSS staff will then review, refine, and finalize before sending to complainants. A source from the FSS elaborated on the efficiency gains, stating, “For insurance-related complaints alone, which number in the dozens, this concept creates a kind of ‘standard response template.’”
While the FSS plans to commence full-scale generative AI development in late next year following preparatory work, the agency is acutely aware of potential challenges. A primary concern is the risk of the AI developing biases by learning from historical data that might have favored financial institutions. The FSS has explicitly addressed this in its development plan, emphasizing the need to “present measures to minimize bias and distortion risks—such as those arising from data training favorable to financial companies—and ensure fairness.”
Also Read:
- Artificial Intelligence Reshaping Financial Services: A New Era for Risk, Fraud, and Investment
- Finland’s Central Bank and FIN-FSA Host Seminar on AI’s Transformative Role in Economy, Finance, and Supervision
Beyond complaint handling, the FSS intends to extend the application of generative AI to other critical areas, including systems designed to detect unfair trading practices, enabling immediate identification of suspicious transactions.
