Generative AI in Public Opinion Guidance during Emergency Public Events: Challenges, Opportunities, and Ethical Considerations

The importance of effective public opinion guidance during emergency public events has been underscored by the COVID-19 pandemic. Generative AI models, particularly GPT, have demonstrated potential in various fields, including journalism and information science. This study investigates the effectiveness of GPT-generated content in shaping public opinion during emergency public events and the factors influencing its impact. We conduct a quantitative content analysis and a qualitative case study to analyze the challenges and opportunities of using GPT during emergencies and provide best practices for its use in public opinion guidance. Our findings reveal the potential of GPT in providing accurate and timely information, countering misinformation, and promoting effective responses. However, limitations such as biased datasets, adversarial attacks, and resource constraints must be addressed to ensure responsible and ethical use of GPT in public opinion guidance. We provide recommendations for journalists, public relations professionals, and emergency management officials, while also outlining future research directions.