Improving citizen-government interactions with generative artificial intelligence: Novel human-co...
Enhancing Citizens' Understanding of Government Policies through Artificial Intelligence
Integrating Generative AI and Language Models for Policy Communication
Effective communication of government policies is crucial for transparency and public engagement. However, challenges like accessibility, complexity, and resource constraints hinder this process. The integration of generative AI and artificial intelligence technologies into public administration has significantly enhanced government governance, fostering dynamic interaction between public authorities and citizens.
This paper proposes a system leveraging the Retrieval-Augmented Generation (RAG) technology combined with Large Language Models (LLMs) to improve policy communication. Addressing accessibility, complexity, and engagement challenges, our system utilizes LLMs and a sophisticated retrieval mechanism to generate accurate, comprehensible responses to citizen queries about policies.
System Architecture and Implementation
Our system comprises three main elements: a document retriever that uses semantic search to identify relevant policy documents based on user queries, a document reader that extracts key information from retrieved documents using natural language processing techniques, and a generator that synthesizes information to produce coherent and accurate responses using GPT-3.5-turbo.
The system's architecture facilitates dynamic updating and scaling, ensuring real-time incorporation of new information. Additionally, an external filter layer screens for irrelevant or harmful responses, maintaining the system's integrity and accuracy.
Quantitative Evaluation and Performance Analysis
We evaluated our system using a dataset comprising 100 policy documents from diverse policy areas from the United States and China. Experiments showed high accuracy rates, particularly in handling quantitative information.
While the system demonstrates strong performance, it may benefit from further tuning to accommodate linguistic and structural complexities. Case studies highlight the need for improved document management and system design to address challenges involving document structure and specificity of queries.
Applications and Real-World Impact
The system offers practical applications in enhancing policy communication and citizen engagement. Potential scenarios include providing real-time information during crises, clearly explaining complex policies to improve public understanding, and boosting transparency and accessibility of policy information through multiple platforms.
Conclusion and Future Directions
The integration of RAG and LLMs has significant potential to enhance government-citizen communication. By continuously refining the technology and addressing its limitations, future work will focus on increasing accuracy, applicability, and accessibility to create a more informed and participatory democratic society.