Séminaires des chairs électorales - 27 avril

Visualizing the impacts of climate change using AI

Erick Lachapelle (Université de Montréal)
 Thomas Bergeron (University of Toronto)
Victor Schmidt (MILA)
Alex Hernandez-Garcia (MILA) 
Yoshua Bengio (MILA)

Existing research suggests that climate change is perceived as a spatially and temporally distant threat, prompting researchers to explore various forms of risk communication that better engages the general public. However, results from attempts to proximize the threat of climate change in the literature have been mixed.  In this presentation, we present results from two survey experiments designed to test the potential of a deep learning algorithm to generate personalized images to help citizens visualize the possible effects of climate change where they live. In a first experiment conducted on a sample of 1,099 residents living in the province of Quebec, we demonstrate that exposure to AI generated images of major flood events produce reactions among treated respondents that are no different from respondents exposed to an image of a real flood. In a second pre-registered experiment administered to samples drawn from Canada (n=2000) and in the United States (n=3000), we randomly assign respondents to a treatment asking them to self-administer a personalized AI-generated image of their home with a simulated 1m flood in real time. We found that, relative to the control group, treated respondents showed higher levels of perceived harm from climate change and floods, while also demonstrating higher levels of concern, greater behavioural intentions (e.g., contact their MP or discuss climate change with their family) and policy support (e.g., Increase taxes on fossil fuels). These treatment effects persisted after four weeks. Overall, we argue these results have important implications for thinking about how best to engage citizens with respect to the effects of climate change using personalized images of places people care about.


Contactez Semih Çakır si vous souhaitez assister à la présentation.

Ce contenu a été mis à jour le 25 avril 2022 à 10 h 04 min.