Ask most people to describe artificial intelligence, and a few things likely come to mind: self-driving cars, facial recognition or ChatGPT helping you write that university essay you’ve been putting off.
But did you know that it can also be a valuable tool to prevent deforestation, preserve ecosystems and protect wildlife?
For many, that’s still a novel concept.
“I think it’s important that we’re seeing an increase in people’s interest in how AI can play a role in mitigating and adapting to climate change,” says Jake Okechukwu Effoduh, an assistant law professor at Toronto Metropolitan University and a Climate Reality Leader.
“There are AI tools that do species identification and monitoring, and algorithms that can help analyze images, videos and sounds, even in remote and inaccessible areas.”
While some of that technology is still nascent, he adds, “when it comes to biodiversity, AI helps with habitat mapping, using satellite imagery and machine learning with conservation work, especially in the Global South.”
Much of the value of AI to conservation comes from its ability to take complex, continually evolving data and make it accessible to the people who need it.
For several years already, the UN Food and Agricultural Organization (FAO) has wielded a powerful, open-source platform called SEPAL (System for Earth Observation Data Access, Processing, and Analysis for Land Monitoring).
Funded by Norway’s International Climate and Forests Initiative, SEPAL provides real-time data to users in 64 countries on what’s happening in and around their forests, including deforestation and wildlife poaching.
This means it can send alerts on disturbances to forest rangers in a couple of days rather than months.
“The result is a shift from finding out too late to finding out fast enough to intervene,” FAO senior forestry officers Julian Fox and Anssi Pekkarinen wrote in an op-ed last year.
Then there are initiatives like Harvard University’s Protection Assistant for Wildlife Security (PAWS), a machine learning system that helps rangers patrol protected areas by telling them where they are most likely to find snares and traps set by poachers.
“PAWS uses historical patrol data to make predictions about where poachers will be in the future so that we can help rangers make the best use of their limited resources,” says Lily Xu, a computer scientist specializing in AI and a postdoctoral fellow at the University of Oxford.
“Much of that comes from remote sensing, with satellites that take information about temperature, precipitation and how those change over time,” she adds.
Land cover, elevation, the locations of rivers, and animal distributions can also be monitored. “Then, we build a predictive model using machine learning to figure out the patterns in where snares typically are, how that relates to these landscape features, and use that to decide where to send our rangers,” Xu explains.
Despite their promise, AI conservation tools have yet to be deployed to their full potential, in large part due to a lack of investment.
“To really unlock the potential of these technological solutions, there is a need to provide robust scientific information to those who are in charge of forest management, to improve data collection and to invest in the capacity building of local researchers and communities,” said Damase Khasa, a professor at Université de Laval, at the OFAC Hybrid Forum on Central African forests in June.
While many countries have officially designated areas vulnerable to deforestation as protected, which Xu calls ‘paper parks,’ they have nowhere near enough resources to actually patrol and search for illegal activity.
According to a 2022 study, governments are only investing USD 24.3 billion a year in managing existing protected areas – just a fraction of the USD 67.6 billion needed.
What’s more, Xu points out, community and Indigenous lands receive little to no funding for protection. That means communities with the least resources must take on protection duties themselves.
While PAWS works in Uganda, Belize and Cambodia, other countries are also using its machine learning model to make their own predictions in partnership with the Spatial Monitoring and Reporting Tool (SMART).
A consortium of eight NGOs dedicated to managing protected areas, SMART runs a software system used by park managers and rangers on patrol, who can in turn feed information into the system based on their findings and observations.
“SMART is a widely adopted platform that’s in use in over 1,000 protected areas around the world, and PAWS is one of the plug-in resources available on the platform,” says Xu.
Scientists aren’t alone in reaping the benefits of AI: the technology is also providing farmers with information to help them improve crop yields.
AMINI, a climate tech startup based in Nairobi, Kenya, is building data infrastructure for Africa and the Global South.
“We collect and analyze detailed data from space using our AI machine learning tech,” says its market analyst Chadi Assoualma.
“This enables us to combine different types of data – socioeconomic, geospatial and environmental – to provide insights to insurance companies and other businesses, cooperatives and farmers.”
Based on satellite imagery from NASA and the European Space Agency, these tools monitor vegetation and soil health. AMINI then offers a variety of solutions to support farmers in making data-driven decisions.
These tools also are used in landscape restoration, says Assoualma, “helping people understand the land and find the best places for restoration projects.”
Effoduh has also seen the benefits of AI for farmers and communities in various African countries, including precision farming, water harvesting and climate modeling to predict future climate change scenarios.
“Precision farming has done a lot of change in in terms of helping farmers prioritize what seeds to plant, where to plant, how to plant – and to maximize their yields,” he says.
Likewise, information on water can also help local communities prepare and adapt to drought and water scarcity, he says.
“They’re using AI on what might be considered mundane or very simplistic things, but that’s where the life change is happening,” he adds. “That’s where we are transforming people’s lives.”
Despite its value in conservation, artificial intelligence comes with one big drawback: its massive carbon footprint.
Making its own outsize contribution to climate change, large language models like OpenAI’s ChatGPT use staggering amounts of energy and water.
And according to the International Energy Agency, global data centers are set to double their electricity consumption by 2026.
That means they’ll be using as much electricity per year as Japan – never mind African countries, where, as Effoduh points out, many communities don’t even have electric lighting.
Then there AI’s reliance on cheap labor: the millions of poorly paid data annotators and content moderators who work in often appalling conditions.
“AI tools are exploiting workers from the Global South, including workers in prisons and in refugee camps,” Effoduh points out.
“The promises of AI are equal to its perils and risks,” says Xu. “Energy consumption is a major concern, along with the spread of disinformation, automation and elimination of jobs.”
PAWS uses far less energy than those energy-guzzling large language models, she emphasizes – meaning its benefits still vastly outweigh its costs.
“If I had terabytes of information about where elephants are and where poachers are, that would be amazing,” she says. “But for conservation, we are data-scarce.”
“The efficiency gains in resource allocation is much greater than the electricity that it costs to run these small models,” Xu adds.
“The key difference is that we are trying to develop AI that is effective when we have missing data and incomplete information, whereas the AI that is extremely power-hungry is only effective in situations where we have a lot of information already available.”
In other words, Xu argues, it’s important that we distinguish between the different types of AI tools that exist – and prioritize investing in those making a positive impact on society.
Éliane Ubalijoro, CEO of CIFOR-ICRAF, echoed the sentiment in an op-ed in Forests News last year. “We don’t yet have a great track record of ensuring that technological advances are shared with those who need them most,” she wrote.
“The unique contexts and challenges faced by communities at the forefront of climate risks are often ignored, and algorithmic bias can deepen inequity and reinforce discrimination.”
Effoduh agrees that the ‘AI divide’ is a major obstacle, and for AI to truly be a transformative force for people and the planet, it needs to be inclusive, equitable and harm-free.
“We need to ensure that the people who need this the most are at the table when the technology is designed and when it is deployed, and that they are also at the table to share the benefits of what this technology can do,” he says.
Effoduh believes big tech and the wider private sector need to do much more. “They’re the ones who have the money, the investment. They do a lot of product development. They have access to open data.
And while many companies say they’re working towards reducing their emissions, the facts often paint a different picture.
“When we break it down, we are not seeing any considerable change.”
Finally…
…thank you for reading this story. Our mission is to make them freely accessible to everyone, no matter where they are.
We believe that lasting and impactful change starts with changing the way people think. That’s why we amplify the diverse voices the world needs to hear – from local restoration leaders to Indigenous communities and women who lead the way.
By supporting us, not only are you supporting the world’s largest knowledge-led platform devoted to sustainable and inclusive landscapes, but you’re also becoming a vital part of a global movement that’s working tirelessly to create a healthier world for us all.
Every donation counts – no matter the amount. Thank you for being a part of our mission.
