Seeing the World Submerged: Exploring the Potential of AI Image Generation for Climate Change Communication

Typical climate change imagery falls short in bringing climate change close to the reality of people's lives, oftentimes failing to have an impact on pro-environmental attitudes and behavior. While this highlights the need for innovative approaches to climate change communication, the rapid growth of artificial intelligence (AI) technologies holds great promise for addressing societal challenges such as climate change. In the present research, we therefore explore how AI image generation can be used as a novel means for climate change communication. Specifically, we focus on humans’ ability to co-create with generative AI in order to produce highly tailored visualizations of climate-change-related consequences in familiar places. Thirteen participants were instructed to choose either a place of personal significance or a neutral place (high vs. low place attachment) and use AI image generation to visualize these places as flooded scenarios. We captured their experiences through in-depth semi-structured interviews while recording their image production process. We analyzed the data using qualitative and quantitative measures. Creating the visualizations elicited multifaceted effects on participants, including emotional arousal, behavioral intentions, and deep contemplation. Our reflexive thematic analysis identified several important psychological concepts that played a central role in shaping these responses: Psychological distance, place attachment, and self-efficacy. The results underscore the extraordinary potential of AI technologies to enhance relevance to individuals by personalizing climate change communication. However, the desired effects of AI image generation have certain limitations, emphasizing the importance of responsible use. Therefore, we present nine recommendations to provide practical guidance and direction for further research in this emerging field.