Guelph professor talks about the use of AI in conservation

The use of AI in conservation efforts is a topic Eric Nost, an associate professor of political ecology at the University of Guelph, analyzes.

“Say, even 15 years ago, if I wanted to know how land cover had changed in my watershed. I would need the right expertise on my team. I would need a researcher with access to specialized software to do that kind of analysis. I might even have to purchase the data,” said Nost.

In the past, the limited number of satellites and the time spent on analysis meant the waiting period to just receive the data could take anywhere from weeks to months. Now, AI can have the data ready with same-day delivery.

“With something like Global Nature Watch, the analysis is as simple as entering a few prompts, which is not only convenient but shortens the kinds of the time needed to do that work,” said Nost.

Global Nature Watch is an open online platform that uses satellite data, cloud computing, and AI to monitor forests and natural ecosystems across the planet in near real time. Researchers use discriminative AI, which categorizes information to analyze images and identify wildlife, plants, and patterns.

To understand the landscape of how conservation efforts use technology, Nost conducted a 2023 survey examining how 45 conservation organizations across Canada and the United States use it.

“The first thing it showed us is that smaller conservation organizations are still able to use a lot of these emerging technologies. There are big questions around capacity and cost, because a lot of these technologies can be quite technical to use, or require certain kinds of expertise, or you may even have to pay money to access the data or the software,” said Nost.

Beyond these questions, Nost said a bigger dilemma for those working in conservation is the environmental impacts of using AI.

“It is a bit of a paradox in that, you know, if I’m working at a conservation organisation or if I’m a citizen conservationist, I’m doing work to advance species protection, habitat protection, and increasingly a lot of that has to do with mitigating climate change, and here we have something that, in many ways is making the problems of climate change and other problems such as water availability potentially worse,” said Nost.

Using AI requires resources such as water and land for data centers and metals mined for computational chips that perform the calculations to develop AI models.

“Most of the biggest hyper scaled data centres that are really focused on AI are in the U.S. So, they get those chips sent to the U.S. to be housed in data centres- first of all, you need land for a data centre. These are increasingly enormous warehouses that have a significant footprint, so that’s taking land that could otherwise be farmed or be species habitat or whatever it currently was, and paving it over and building something on top of it.”

On top of the land used for data centers, Nost noted that data centers are major electricity consumers.

“There are proposals [for data centers] that would generate as much electricity as the entire state of Utah, or seven times the city of New Orleans, so the magnitudes here are significant,” stressed Nost.

“Then you need water for these data centres. So, the water is for two things: the water is used at the power plants to produce steam, but then the water is also used at the data centres themselves to cool the chips. Computation requires work; work produces heat, and you don’t want your chips to overheat because then they won’t work.”

Traditionally, water has been used to cool the chips. While newer data centres are being proposed to use different cooling systems, the trade-off is that more energy is required for cooling.

“So, we might solve some of the water usage issues, but we might be making some of the energy usage issues worse. And then once you actually turn the data centre on, you have some other issues, including things conservationists would be concerned about, like noise. These are industrial facilities. They are on 24/7. They’re humming. They’re buzzing, and they produce sound at frequencies that humans can’t hear as well, and that can be of concern to animals and even plants,” said Nost.

While AI is helping to speed up the data collection process for conservation, the environmental consequences of its use are something Nost asks everyone in conservation to consider right now.

“Like with anything, like with driving, for instance, there’s trade-offs, and we live in a world where we kind of have fairly limited choices. Our choices are structured for us, so increasingly It might be harder to make choices to not use AI, despite how that might conflict with some of our ethical priorities. But this is very much in flux right now, where people and organisations are trying to figure out how they should or shouldn’t use AI and how it aligns with their ethical values.”