Hope versus Despair in AI-generated Climate Change Visuals

Abstract

This paper investigates the paradox of Generative Artificial Intelligence (GenAI) in climate communication, examining its capacity to encode and propagate visual narratives of hope versus despair regarding climate futures. The climate crisis is framed as a hyperobject, whose scale and inaccessibility have historically resisted effective visual representation, often resulting in public disengagement, increased Psychological Distance, and reduced Self-Efficacy. While GenAI promises personalized, localized risk visualization, this study reveals that it risks amplifying existing cultural biases, leading to a pervasive dystopian default that constrains the collective imagination. Employing a qualitative, arts-based research design, we conducted a visual analysis of a corpus of 285 AI-generated images to deconstruct the aesthetic conventions and compositional strategy of GenAI output. Our findings show that GenAI's dystopian cinematic tropes rely on an iconography of ruin and cool, subdued tones that foster apocalypse fatigue and affective paralysis. Conversely, utopian outputs default to a techno-solutionist aesthetic, characterized by vibrant greens and an exaggerated focus on renewable infrastructure, which risks creating false hope by simplifying the necessary systemic reforms.

Ultimately, the paper argues that GenAI’s greatest utility resides in its capacity to generate hybrid climate imagery. This can be achieved by blending signs of ruin with localized human action, grounded psychologically in the sublime, thereby compelling a more nuanced, self-reflective engagement with the crisis. The study concludes by advocating for the cultivation of Critical Hope through action-oriented visualizations that leverages episodic despair as a corrective to naive optimism, facilitating collective agency and persistence amid the uncertainty of the Anthropocene.