Optimizing Natural Language Processing Using ChatGPT for Automated Question and Answer Systems in Public Services

Abstract

The use of artificial intelligence (AI) technology is growing rapidly in various sectors, including public services. One of the main challenges in digital public services is how to provide answers that are fast, accurate, and easy for the public to understand. Natural Language Processing (NLP) is key in the development of automated question-and-answer systems. This study evaluates and optimizes the use of ChatGPT as an NLP model in a question-and-answer-based public service system. The methods used include literature review, system design, prototype implementation, and performance evaluation based on accuracy, response speed, and user satisfaction. The results show that ChatGPT can increase answer accuracy by up to 88%, reduce the average response time to 2.9 seconds, and increase user satisfaction to 85%. The system is also able to adapt its language style to make it more understandable to the public. These findings indicate that ChatGPT can be effectively integrated into digital public services, with important considerations regarding ethics, data security, and bias mitigation.