Advertisement

Montreal-made website uses AI to depict impacts of climate change anywhere in the world

Click to play video: 'Ahead of COP26, thousands march for climate action in Brussels'
Ahead of COP26, thousands march for climate action in Brussels
Some 25,000 people took to the streets of Brussels in Belgium on Sunday to call for climate action ahead of the U.N. Climate Change Conference (COP26), based on numbers provided to local police – Oct 10, 2021

Imagine Parliament Hill blanketed in orange skies, floodwaters climbing the sides of BC Place stadium or a thick layer of smog fogging the view of Halifax from Citadel Hill.

These are all scenes depicted on a website published Thursday that blends artificial intelligence with geography to show the potential impact of climate change on almost any address on the planet.

The website, thisclimatedoesnotexist.com, was created by Mila, a machine-learning and technology research institute in Quebec founded by Yoshua Benigo, one of the godfathers of AI.

The website can apply filters showing the impacts of flooding, wildfires or smog to any address available through Google Street View and is meant to raises awareness about future scenarios that could arise if the world’s response to climate change continues to fall short.

Read more: Climate change is ‘biggest health threat facing humanity,’ WHO says

Read next: West Edmonton Mall closes Mindbender indoor roller-coaster

Story continues below advertisement

Victor Schmidt, a lead researcher on the project, said the site is not meant to predict climate change, but rather, it strives to make the issue more personal for people who can’t visualize something that is seemingly far off or something that could impact another community sooner than theirs.

“Just because it may not happen to them, doesn’t mean it won’t happen to other people elsewhere or sometime in the future,” Schmidt said.

To boost empathy, the website prompts people to search their current or childhood home, workplaces, favourite restaurants and travel destinations.

The images the website will return are built around generative adversarial networks, or GANs, a class of machine-learning frameworks designed by Mila that allow a computer to create and transform images.

Read more: Will Canada see a repeat record-setting heat wave in summer 2022?

Read next: Marilyn Manson accused of raping underage girl in new lawsuit

The site doesn’t adjust water levels or air quality conditions based on where an address is, so users won’t find coastal areas completely underwater or wildfire-ridden neighbours more engulfed by dusty skies than less fire prone regions.

This is intentional, Schmidt said. Mila didn’t want people visiting the website, discovering their neighbourhood was facing less distressing climate change impacts than others and then deciding not to care about the issue.

Story continues below advertisement

“It’s about connecting with other people and trying to make common sense feel closer,” he said.

“Making climate change more personal is going to try and help bridge that gap.”

Read more: Should climate change action cost the economy? Canadians unsure, poll says

Read next: Parks Canada updating its reservation system to book camping and other activities

Schmidt hopes people will visit the website built with the support of National Geographic Society, Microsoft, BCG Gamma and Borealis AI, and leave feeling more aware about climate change and its affects.

Research on the Canadian Centre for Climate Services website shows climate change has already caused the country’s average annual temperature to rise by 1.7 C from 1948 to 2016, about double the global rate.

In Northern Canada alone, the average annual temperate has risen by 2.3 C, about triple the global rate, over the same period.

Almost simultaneously, the summer sea ice area in Northern Canadian waters decreased by nearly 7 per cent per decade on average between 1968 and 2015.

It is expected to keep decreasing to the point where Arctic waters could be nearly ice-free by the 2050s.

Sponsored content