Demonstration of AI Human for Handling Civil Petitions in National Pension Lump-sum Payments

Overview

As public interest in the National Pension grows, the volume of related civil complaints has surged, leading to increased workloads and a shortage of response personnel. In particular, the high frequency of inquiries regarding lump-sum pension payments requires rapid and accurate responses. To address this, the project aims to demonstrate a complaint response system by introducing an AI Human-based concierge service capable of responding in a manner similar to actual staff.

Project description

To build the AI Human-based response service for lump-sum pension inquiries, various data were collected and training models were developed. Over 1,000 sets of data, including the National Pension Service's (NPS) guidelines, official documents, guides, and Q&A records, were utilized.

The model development process followed a structured sequence: 1. Data preprocessing and dataset construction, 2. AI model integration and fine-tuning, 3. Development of chatbot functions for lump-sum payment inquiries, and 4. Service testing in connection with the AI Human interface.

Key technological features

• SmartChat-based Query Generation: Conducted data preprocessing, indexing, and generated Q-A pairs based on SmartChat
• SmartChat: Implemented a hybrid approach combining the strengths of SearchGPT and SmartChat for both search and chatbot solutions

Expected outcomes

The introduction of the AI Human-based system will enable more accurate and prompt responses to inquiries regarding lump-sum pension payments. This is expected to reduce the burden on personnel and increase citizen satisfaction. Furthermore, by supporting keyword and query-based searches, the system allows users to find desired information quickly, providing an environment where pension-related complaints are handled effectively through intelligent search and chatbot functions. Based on the Proof of Concept (PoC) results, future plans include additional model training to expand the service beyond pension inquiries to various other administrative services.

Launch year: 2023

Attribution

Case study courtesy of the National Information Society Agency, Republic of Korea, Apolitical's Content Partner. Sourced from Use Cases of Public AI Service, Vol.2, https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do?cbIdx=31975&bcIdx=29182&parentSeq=29201