ARI SHAPIRO, HOST: A failing National Park in Malawi has become ground zero for an experiment in conservation. Officials are using artificial intelligence and machine learning to stay one step ahead of poachers and keep elephants from killing villagers. NPR’s Dina Temple-Raston reports from Malawi as part of the NPR series “I’ll Be Seeing You” about the technologies that watch us.
DINA TEMPLE-RASTON, BYLINE: The best way to get close to an elephant is to use an old-fashioned tracking method. We saw the fresh dung piles along the road and followed those and found him in this clearing. An enormous elephant is munching on grass less than 20 feet in front of us. It’s magical. Liwonde National Park is 200 square miles of scenes just like that, and seeing them makes it hard to believe that just four years ago, the park was on the verge of collapse.
CRAIG REID: So I always describe Liwonde as we found it as being in a state of terminal decline.
TEMPLE-RASTON: That’s park manager Craig Reid. He arrived here four years ago with a nonprofit organization called African Parks, and his job was to bring Liwonde back from the brink.
REID: So effectively, what would have happened had we not intervened would be a total elimination of all wildlife over the 10 year period following.
TEMPLE-RASTON: The things that go wrong in a failed park go wrong very fast. In Liwonde’s case, it went beyond crumbling infrastructure or washed-out roads. Poaching was endemic, and elephants were killing villagers outside the park. Various Donzani has a small house just outside the park’s fence line. He’s a retired schoolteacher.
VARIOUS DONZANI: My house is 50 meters from the fence.
TEMPLE-RASTON: Fifty meters from the fence?
DONZANI: Fifty meters.
TEMPLE-RASTON: Donzani is a subsistence farmer, and he and his family rely on the garden for food. He grows mangoes and corn and rice, which happen to be three of an elephant’s favorite foods. Elephants have a sweet tooth. So proximity, food and opportunity – that’s all you need to spark a human-elephant conflict.
TEMPLE-RASTON: This is the sound from a video of an elephant charging in Liwonde. After hippos, elephants kill more people in Africa than any other animal. Try shooing an elephant off a field, and this is what you face.
TEMPLE-RASTON: An adult female weighs over three tons. A bull elephant is typically over six. Park manager Craig Reid says being killed by an elephant is a horrifying thing.
REID: These big bulls will charge a person, and when people are caught by an elephant, they’re normally using their trunk to knock them on the ground and then kneel on them, crushing them either with their knees or with the base of their trunk, which, at that point, is all curled up and almost like a fist.
TEMPLE-RASTON: Donzani has seen it happen. He said several years ago, a herd of elephants came out of the park and into the fields.
DONZANI: I tell you, it was a disaster. Seven people were killed that day.
TEMPLE-RASTON: Seven people killed in one day. These kinds of deadly episodes have a ripple effect. Smugglers and international cartels will look for local farmers – not Donzani – who have had these kinds of run-ins with elephants, and they’ll promise them enough money to buy food for their families. All they have to do is help the poachers locate an elephant. On the black market, a single elephant tusk can sell for hundreds of thousands of dollars.
UNIDENTIFIED REPORTER: A new study, the Great Elephant Census, suggests a failure to protect the world’s largest land mammals, elephants.
TEMPLE-RASTON: It so happens that around the time Liwonde Park manager Craig Reid arrived in Malawi, one of the founders of Microsoft, Paul Allen, was putting the finishing touches on something he called the Great Elephant Census. It was a massive, multi-million dollar project aimed at counting all the savanna elephants in Africa, and it found that in the seven years between 2007 and 2014, nearly a third of Africa’s elephants were killed, mainly by poachers. Allen wanted a high-tech solution, so he funded a company called Vulcan, and they created a program called EarthRanger. And what it does is it takes all those things park managers know from years of experience and all that paperwork park rangers fill out and puts it through a machine learning program to find patterns and to learn the rhythms of the park.
LAWRENCE MUNRO: Message public (unintelligible).
TEMPLE-RASTON: Liwonde’s operations manager is a man named Lawrence Munro, and part of his job is to plan and schedule ranger patrols. He’s been using EarthRanger to help him deploy those rangers more efficiently.
MUNRO: You can kill the (unintelligible) for this circle.
TEMPLE-RASTON: Munro was standing in front of an image of the park up on a flat screen. It shows the river and clearings and forests. There are little elephant icons tracking GPS location signals from collared elephants, and little rhino icons move across the screen as well. Munro starts clicking a mouse through various EarthRanger menus. He moves through an animation of suspicious activity in the park. You can see a concentration of snares, little icons that look like little Western lassos, and there are a bunch of them right off the main road.
MUNRO: So you can see those snares are up there just to the post.
TEMPLE-RASTON: Munro and the park’s head of security Paul Chidyera decide to set up new checkpoints right where EarthRanger recorded a large number of snares. Before EarthRanger, Munro used to do this kind of analysis with a map and magnets.
MUNRO: You have to rely a lot more on memory, a lot more on radio traffic. Here, you can do a lot more pre-emptive stuff because you can see the picture.
TEMPLE-RASTON: Munro said it used to take a month to start to see patterns emerging. Now he can do that daily.
MUNRO: We study it intently every Wednesday because we want to deploy our guys accordingly, but you can do it daily. It’s basically the speed at which you can strategize.
TEMPLE-RASTON: Strategize and respond. Earlier this year, the head of law enforcement at Liwonde, Paul Chidyera, got an alert on his cell phone after an EarthRanger sensor was tripped. He set up a poacher cam in the area to try to get a beat on what it was. The poacher cam has a special algorithm inside that helps the camera determine whether something going past has a human shape or an animal one. If it looks human, it snaps a picture and sends it with GPS coordinates to EarthRanger and staff cell phones. The poacher cam got a photo of someone coming in.
PAUL CHIDYERA: When he was arrested, he was confronted, and he revealed all of what he has been doing.
TEMPLE-RASTON: In fact, as they scrolled back through EarthRanger’s database, they found other pictures of him coming into the park with weapons. Those pictures were submitted as evidence in his trial. In the end, the poacher was sentenced to 27 years as a repeat offender.
In the past two years, poaching in Liwonde has plummeted. The park hasn’t lost a single high-value animal in 30 months, and those deadly unscheduled elephant trips outside the park – they’re way down, too. While all of this has been going on, EarthRanger’s machine learning algorithm has been training. The program will use game theory and behavioral analysis to tell people like Munro where poachers will go before they get there, allowing rangers to intercept bad guys pre-bang. Poachers will go to where they believe the elephants will be, and they’ll find rangers waiting there for them instead. Pawan Nrisimha of Vulcan, who helped create EarthRanger, says AI can take a long time to learn.
PAWAN NRISIMHA: The biggest difficulty here is really gathering that data to train up that artificial intelligence model to know what to predict.
TEMPLE-RASTON: Now that EarthRanger has been deployed in Liwonde for two years, it’s getting close to having enough data to start making predictions. Nrisimha sees EarthRanger getting poachers before they can even fire a shot in the next year or two.