Skip To Main Content
An image of the globe superimposed on parched earth
Data and technology

Is AI sustainable?

The complex environmental impact of artificial intelligence has grown past ethics in usage debate due to its role in climate change

We’re all aware of the very real and very devastating effects of climate change; we’re seeing extreme weather events, worsening air quality, food shortages and an increasing number of animals facing extinction, to name a few. And while it’s easy to play the blame game—we’re looking at you gas-guzzling vehicles, one-day delivery services and fast-fashion trends—there are many other issues at play. 

In January, the International Energy Agency (IEA) released its predictions for global energy use over the next two years. For the first time, the study looked at the amount of electricity needed to run and sustain artificial intelligence (AI), cryptocurrency and data centres. It’s easy to overlook these innocuous—and invisible—processes. They are, for the most part, designed to help run our lives more efficiently. However, their digital carbon footprint is substantive.

Read More

The IEA anticipates that electricity consumption from these sectors will double by 2026 to around 1000 terawatts. That’s roughly equivalent to the amount of energy used by the entire country of Japan. Shocking? Sure. Catastrophic? Well, the answer to that isn’t as clear-cut.  

“There are many different ways that AI and the climate intersect in both positive and negative ways,” says David Rolnick, an assistant professor and Canada CIFAR AI chair at McGill University, as well as the co-founder and chair of the global nonprofit Climate Change AI. And, he emphasizes, the conversation is far more nuanced than whether or not ChatGPT is good or bad for the environment.  

One of Rolnick’s main areas of focus uses AI to build climate action to optimize electrical grids, map agriculture and monitor biodiversity. And many others in the public and private sectors are using AI at scale to fight climate change. On the other hand, the oil and gas industry is also using machine intelligence to increase production, so much so that it’s estimated to gain an additional $425 billion in profit by 2025. But let’s take a step back for a minute. 

ChatGPT is all the hype right now, but there are many more everyday applications. Using facial recognition to unlock your phone, asking Alexa about the weather or browsing social media are all examples.  

If you’re wondering what all of this has to do with climate change, it’s the large amounts of energy needed to sustain the technology. For most of us, our smartphone or laptop can house AI sufficient enough to autocorrect our spelling or remind us to attach a file to our emails. But if you’re asking it to draft an email or write code, it will need to rely on bigger processors that are housed in data centres.  

These data centres require large amounts of electricity to run and water to cool down. There’s also the intense amount of “training” required to teach these models how to “think.” Because there are so many different types of AI that have different energy requirements, it can be a bit like comparing apples to oranges.  

In April, Canadian Prime Minister Justin Trudeau announced that 2.4 billion dollars of this year’s budget will be allocated to AI, including research, start-ups, infrastructure and job creation, among others. However, no word was mentioned about the environmental effects of these investments and what actions are being taken to implement them sustainably.   

“As the ability to measure the greenhouse gas emissions of AI becomes easier, I’d hope to see an assessment of when it’s appropriate to train brand new models versus adjusting previous ones,” says Taryn Abate, director of research and thought leadership at CPA Canada. “It’s also important to identify purposes for its use that are doing more good than harm.”  

There is a time and place for generative AI, but it’s up to us to recognize when to tap into it. ChatGPT is shiny and new, and as such, it’s being marketed as the solution, whether or not it’s the best tool for the job. Rolnick believes that Google is still the better—and more energy-friendly—search engine for the job. 

“AI is a business, and there’s a big sales pitch going on right now,” he explains. “You need to see beyond the salesmanship…because often the best way to achieve is through something that is actually much more computationally efficient.”