Séminaire du mardi - 19 janvier 2020

Mardi 21 janvier 2020 : Érick Lachapelle et Thomas Bergeron (Université de Montréal). « Visualizing Climate Change with AI ». Pavillon Lionel-Groulx, Université de Montréal, C-4145, 12h-13h.


Résumé (en anglais) Despite expert warnings of the catastrophic effects of runaway climate change and growing public acceptance of climate science regarding a rapidly  warming planet and its human causes, the general public remains relatively unconcerned with this problem. Cross-national research has consistently shown that many people perceive the risks of climate change to be spatially and temporally distant, affecting others, but posing limited risk to them personally. In an effort to build more public support for climate action, both academics and activists have attempted to “proximize” the issue by making climate change more personally salient, with varying levels of success. In this study, we assess how the visualization of the effects of climate change on subject’s places of interest (e.g. their home, their neighborhood) will affect their perception and engagement toward mitigation. We use a Multimodal Unsupervised Image to Image Translation (MUNIT) and a Generative Adversarial Network (GAN) model to develop images of climate-induced floods. This machine learning model will produce a realistic image of recent flooding estimates for their house, neighborhood and places of interest (e.g. metro station; community library). Our objective is to test the effectiveness of this decision-aid technology in helping individuals visualize the effects of climate change, as well as the effects of this information-updating on subsequent emotional states, risk perceptions, policy preferences, behavioural responses, and political engagement with the issue.


Ce contenu a été mis à jour le 30 avril 2020 à 7 h 43 min.


Laisser un commentaire