AI is increasingly used to fight climate change, but it has its own emissions problem


On a farm in St. Peters Bay, Prince Edward Island, a four-wheeled black rover with two outstretched arms moves through a row of thigh-high green leaves, its giant tires kicking up the earth red from a potato field. It seems he’s more comfortable in a red, dusty Martian landscape than on a farm.

“A few people actually stopped on the road to see what was going on,” said Aitazaz Farooque, interim associate dean of the school of climate change and adaptation at the University of the Island. of Prince Edward Island (UPEI).

Meet AgriRobot, a robot trained using artificial intelligence to identify diseases in potato plants.

Farooque leads a team of researchers at UPEI (in partnership with the governments of Prince Edward Island and New Brunswick) who are using AI in new and innovative ways. The AgriRobot is the brainchild of Charan Preet Singh, a master’s student in the university’s department of sustainable design engineering.

“It will generate a map with the location information so that even if someone needs to come in, they don’t need to be trained… they can load that map onto their cell phone,” Farooque said. “It will tell you where these infected plants are and eliminate them.”

Three people stand in a row in a potato field in Prince Edward Island with a four-wheeled robot robot.
The AgriRobot works in a potato field near St. Peters Bay, Prince Edward Island. The robot, created by a UPEI master’s student, was trained using AI to identify diseases in potato plants. (University of Prince Edward Island)

As the climate changes, farmers face more challenges than ever. From floods, droughts and disease to warmer temperatures and changes in growing and harvesting seasons, the agricultural sector is changing rapidly, which means farmers – and technology – must constantly keep pace.

But there is an irony: while AI helps adapt and mitigate climate change, it has its own emissions problem. And this phenomenon will only grow as AI is used for more and more applications.

AI consumes a lot of computers – and energy

“AI is being used in all kinds of ways to combat climate action,” said Priya Donti, co-founder and president of Climate Change AI, a global nonprofit that examines the use of AI. AI in the climate field.

“From helping us better predict solar power and wind power on the power grid to helping us better integrate them into power grids… to helping us map things like deforestation and emissions to the “Using global satellite imagery to understand where deforestation or emissions are occurring in real time.”

AI runs on computers – lots of them – housed in data centers around the world. When AI models work, they need electricity. If that electricity comes from a grid that uses fossil fuels, it contributes to emissions.

A blurry person is seen in a room full of computers.
Artificial intelligence may seem invisible to most people, but the computers that run it are housed in data warehouses like this one and require a lot of electricity. If that electricity comes from a grid that uses fossil fuels, it contributes to emissions. (Oleksiy brand/Shutterstock)

At the same time, the computers in these data centers generate a lot of heat and need to be cooled, which often requires even more electricity.

“Running AI is like running any other computer program. You have input, you want output,” said Yacine Jernite, a researcher in New York who works for Hugging Face, a company that hosts platforms open source where AI models are shared.

“It’s going to perform a lot of operations. And doing a lot of operations for a single response means that the computer that’s performing those operations uses a lot of power and electricity.”

The problem is that no one really knows how much AI contributes to emissions from these data centers.

“We really need to pay attention to the growing emissions footprint of AI,” said Donti, based in Cambridge, Massachusetts.

“And basically one thing that’s hard to understand is that there’s not enough transparency between data center providers, between machine learning entities that are actually creating machine learning algorithms in terms of monitoring and measuring these greenhouse gas emissions.”

WATCH | Using artificial intelligence to predict the weather:

Can AI predict the weather?

An AI weather model from Google DeepMind called GraphCast has outperformed conventional weather forecasting methods by forecasting global weather conditions up to 10 days in advance. Our science and climate expert, Darius Mahdavi, tells us more about its implications for future weather forecasts.

Predicting wildfires before they happen

As we face a continuing climate crisis, scientists are trying to find ways to help us manage the consequences.

In southern countries, locust invasions are increasing, threatening food security. A new tool called Kuzi helps farmers by providing real-time data using satellites, on soil moisture, surface temperature, humidity and more, to predict potential disease outbreaks. It can then send a notification to farmers on their mobile phones.

And as the risk of wildfires increases, engineers and scientists are developing new tools to detect and even predict when they will start.

Dryad Networks, a Germany-based company, has developed solar-powered sensors that can detect a fire before a flame even breaks out.

A small green plastic sensor with solar panels sits on a tree.
This solar-powered sensor, developed by German company Dryad Networks, can detect a fire even before a flame forms. The company has already deployed 20,000 worldwide. (Dryad Networks)

“Behind (the) membrane…is a gas sensor sensitive to hydrogen, carbon monoxide and volatile organic compounds,” said Carsten Brinkschulte, CEO of the company. “So it’s like an electronic nose that can actually smell a fire. And that’s where AI comes in: we run the AI ​​in the sensor, to make it recognize pre-defined machine learning patterns. who have been trained for the sensor. smell of fire.”

The company has already deployed 20,000 worldwide, with a pilot project in part of the forests of California. Brinkschulte said Dryad is also starting a pilot project with an unnamed organization.

AI has a cost for the environment

AI has huge potential, experts say, but first we need to have a better idea of ​​its contribution to emissions – and the transition to renewable energy.

“We both need to make the grid greener, and we need to make serious choices about how to make AI models more effective for the places where we will use them,” said Donti, of Climate Change AI.

“But also, as we must do for each sector, reassess what uses are worth the electricity that arrives.”

And this extends to personal use of AI, as not all uses of AI are the same. Showing him two pictures, one of a dog and one of a cat, and asking him to choose the cat uses much less energy than asking him to create or calculate something.

While you can have fun creating filters for yourself or asking questions of generative AI like ChatGPT, it comes at a cost in terms of emissions. In fact, one study suggests that every time AI generates an image, it is use enough power to charge a cell phone.

“We certainly shouldn’t think of AI as a costless thing,” Donti said. “I think it’s very easy to visualize with this abstract thing on your computer that has no impact, but it does.”

WATCH | Inside the University of Prince Edward Island Climate Change Lab:

A tour of the new climate change laboratory in St. Peter’s Bay

From drones to dormitories, the state-of-the-art St. Peter’s Bay Research Center will host world-class students and researchers studying the many facets of climate change.

Source link

Scroll to Top