3.08 Case study: Generative AI for climate visualisation

One of the greatest challenges in mobilising climate action is the psychological distance of the threat; for many, concepts like rising sea levels or severe drought remain abstract and remote. This case study examines how generative AI is being used to bridge this gap by creating tangible, personal, and emotionally resonant visualisations of potential climate futures.

The “This Climate Does Not Exist“ project, developed by researchers including those at Mila in Montreal, is a pioneering example. It uses a type of generative AI called a Generative Adversarial Network (GAN) to create photorealistic images of what a user’s own home, street, or other familiar location would look like under the impact of floods, wildfires, or smog. A user can input any address, and the AI generates a stark “after” image. The project’s goal is not to provide a precise scientific forecast but to trigger an emotional reaction, to make the abstract threat of climate change feel immediate and personal, thereby enhancing public engagement and inciting action.

A similar approach was used by the UNDP Accelerator Lab in Panama, which conducted a workshop with university students using generative AI to visualise their neighbourhoods in both dystopian and utopian climate futures. For the dystopian scenario, students used prompts like ‘drought’ and ‘flood’, witnessing familiar places transform into scenes of disaster. For the utopian scenario, they used prompts like “nature-based solutions” and “bioclimatic architecture,” producing inspiring images of a resilient and sustainable Panama.

To show the subtitles in your language, please click the gear (settings) icon, then click on Subtitles/CC, then Auto-translate and pick your language from the list.

These exercises yielded powerful outcomes. They sparked critical discussions about climate inequality, as students observed that wealthier areas were often less affected in the generated flood scenarios. They also revealed the limitations of the technology, reinforcing the critical need for the “human-in-the-loop.” In one instance, a student noted that the AI had designed a beautiful garden for a utopian Panama, but filled it with plants that could never grow in the local climate. This highlights that while generative AI is a powerful tool for sparking imagination and conversation, its outputs must be grounded and validated by local knowledge and human expertise to be truly meaningful and effective.