Implementation and evaluation of AI-based Citizen Question-Answer Recommender (ACQAR) to enhance citizen service delivery in Singapore public sector: A case study

• Main Author: LEE, Hui Shan
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2024
• Subjects: applied AI; Digital Government; service science; generative AI; citizen services; Artificial Intelligence and Robotics; Computer Sciences
• Online Access: https://ink.library.smu.edu.sg/etd_coll/562; https://ink.library.smu.edu.sg/context/etd_coll/article/1560/viewcontent/GPEN_AY2020_EngD_Alvina_Lee_Hui_Shan.pdf
• Institution: Singapore Management University

Government agencies prioritize citizen service delivery to foster trust with the public. Technological advancements, particularly in Artificial Intelligence (AI), hold promise for improving service provision and aligning government operations with citizens' needs. Yet the inherent inflexibility of Service Level Agreements (SLAs) often overlooks the nuances of human emotions and the varied nature of citizen inquiries, exacerbated by a lack of tools to guide appropriate responses. This dissertation aims to address the gaps of overlook of human emotions and non-support for appropriate responses, by exploring the following questions: (1) Can a predictive model incorporating both numeric and textual data effectively forecast SLAs? (2) How does emotion analysis impact the predictive model's efficacy? (3) Does integrating a question-answer recommender, augmented with ChatGPT, improve citizen satisfaction and the efficiency of customer service officers? To investigate these questions, a final pilot system known as AI Based Citizen Question-Answer Recommender (ACQAR) was developed, employing techniques such as Latent Dirichlet Allocation (LDA), Logistic Regression, use of Empath Library, and ChatGPT. This dissertation further dive into the AI Based Citizen Question-Answer Recommender (ACQAR) system's implementation within a Singaporean government agency to enhance service delivery. ACQAR attempts to generate contextually aware responses for customer service officers. The study aims to optimize government-citizen interactions in the digital age, where citizens expect efficient, personalized, and empathetic services. The findings of this pilot system shed light on the potential of the AI-based Citizen Question-Answer Recommender (ACQAR) in improving the efficiency of Citizen Service Officers (CSOs) in government agencies. The pilot trial revealed a notable decrease in average resolution time for CSOs after the implementation of ACQAR, suggesting enhanced responsiveness in addressing citizen inquiries. Additionally, the post-service survey data indicated an improvement in citizen satisfaction, particularly in the understanding of concerns and the overall experience. The study's contributions lie in its novel approach to bridging the gap between SLAs and human emotions in citizen inquiries, shedding light on the potential of AI integration in government service delivery to deliver not only prompt responses, but also appropriate replies. It further offers insights into the practical challenges and implications of AI adoption, proposing strategies for smoother integration and risk mitigation within government agencies. It is to note that the content of this dissertation is organised with the identification of gaps via the literature review in Chapter 2. This chapter examines AI's evolution in service delivery, emphasizing its potential to transform government services. Research gaps such as traditional use of SLA was unable to detect human emotions in citizen inquiries and guide appropriate responses, lack of AI readiness framework and AI Explainability (XAI) for government agencies were identified. Chapter 3 presents the case study background. Chapter 4 depicts the first version of the pilot system built which was known as Empath X SLA predictor. Findings showed that the inclusion of human-centric indicators to represent human emotions in the prediction of SLAs does not affect accuracy, in fact it introduces texture to such traditional indicator of citizen delivery standards. Chapter 5 shares about the second version of the pilot system built which was known as Citizen Question Answer System (CQAS). While the accuracy of this second pilot system was not optimal, it sheds light on areas of improvement for such a system, leading to the eventual successful build of ACQAR in Chapter 6. Chapter 6 outlines ACQAR's design and implementation, including integration details and pilot implementation within the Singaporean government agency. Findings revealed that the pilot system does shorten the time taken by a customer service officer to respond to the citizens’ inquiries, while improving citizen satisfaction rates. However, the implementation also revealed that a framework to improve explainable AI (XAI) is required. Considering the challenges in AI adoption highlighted previously, Chapter 7 assesses a government agency’s readiness for AI adoption and proposes a framework for smoother integration and risk mitigation for government agencies. This framework was implemented within the case study and with that, suggested countermeasures were shared. The dissertation concludes with a summary of findings, contributions, limitations, and recommendations for future research and practice. The research done in this dissertation will contribute to understanding AI's role in public administration, offering insights into practical implementation and challenges associated with AI adoption in the public sector.