Abstract
Science communication plays a critical role in bridging the gap between scientific knowledge and the general public, fostering understanding, engagement, and collaboration. With the rise of online digital media and generative artificial intelligence (AI), new opportunities and challenges have emerged in creating, disseminating, and engaging with science content. However, the evolving computing technologies also bring challenges to science communication. The blurring of roles between journalists and scientists, combined with the erosion of professional diversity and authenticity, raises concerns about transparency, proficiency, and trustworthiness in content creation for science communication. Addressing these issues requires careful integration of AI with human expertise, transparent practices, and inclusivity to preserve the integrity and credibility of science communication, aligning closely with the core focuses and principles of human-computer interaction (HCI).To explore how generative AI can support science communication in transiting complicated scientific knowledge to understandable and engaging media forms, this dissertation starts with a large-scale analysis of science knowledge videos on Bilibili. Through the study, we uncovered how communicators strategically leverage visual and narrative elements to engage viewers, revealing the crucial role of emotional arousal and interactive elements in fostering science dialogue. Building on these insights, we further developed RevTogether, a multi-agent system supporting iterative revision of science stories. By employing AI agents to provide feedback from different reader perspectives, RevTogether demonstrates how generative AI can facilitate the creation of accessible yet scientifically accurate narratives. Finally, we designed and implemented SpeechCap, an innovative system in social virtual reality (VR) that enhances interpersonal communication by converting speech into interactive visual elements, showcasing how AI-driven multimodal approaches can enrich science communication in immersive environments. Collectively, these studies contribute to our understanding of computer-mediated science communication in three key aspects: (1) identifying effective strategies for engaging diverse audiences while preserving scientific rigor, (2) demonstrating how AI can support science communicators through personalized feedback and creative assistance, and (3) exploring novel interaction paradigms for science communication in emerging media platforms such as VR. The findings inform the design of future AI-enhanced tools for science communication and highlight the importance of balancing engagement, accuracy, and accessibility in digital science discourse. This research advances the field of HCI by establishing a framework for understanding and exploring potential applications of computational media and generative AI to enhance science communication.
| Date of Award | 15 Aug 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Jiawei MA (Supervisor) & Zhicong LU (External Co-Supervisor) |