Jean Yang:
In films and literature, a wolf’s haunting howl can signify danger or untamed nature. In real life, researchers at Yellowstone National Park analyze these howls using cutting-edge AI technology to better monitor and track wolves. PBS Montana’s Matt Standal explains 653.
Man:
The wolves that have crossed the river are howling.
Matt Standal (voice-over):
Here in Yellowstone National Park’s Lamar Valley, wolves from one of the park’s nine packs have made a killing.
Man:
Let’s go ahead and feel free to take a look over there. This is the Junction Butte pack.
Matt Standal (voice-over):
It’s the end of August, the peak of the bison rut, or mating season, and a time when wolves increasingly prey on bison. Many are injured or weakened in the fierce competition to find a mate, and the wolves take advantage of this.
Jeremy Sundaraj, the park’s wolf technician, monitors the pack as they feed and teaches curious tourists about wolves.
Jeremy Sunderraj, Wolf technician:
So what we’re trying to do here is just count how many there are, record their behavior. If we can see what the carcass is, it is almost certainly a kill simply based on how they behave around it. And if we stay quiet, maybe we can hear them scream.
Matt Standal (voice-over):
Their screams have become the heart of a cutting-edge new conservation project, using artificial intelligence to decode sound recordings. This development in the field of bioacoustics could redefine how such wolves are monitored in the wild.
Man:
You have it.
Matt Standal (voice-over):
Since 1995, when gray wolves were reintroduced to Yellowstone, park biologists have used planes to spot them, helicopters to track them and dart guns to mark them so they can be fitted with radio and GPS collars.
Man:
And you can see that most of the wolves’ conversations take place at night.
Matt Standal (voice-over):
Bioacoustics now offers a new, less invasive way to study them, using sound and advances in artificial intelligence to potentially one day decode wolves’ communication by matching their howls to specific behavior.
Dan Stahler, Senior Wildlife Biologist:
Not only can we hear them here and record their howls 24 hours a day, 365 days a year, but we can often relate wolves’ behaviors by observing them when they vocalize. What is the cause and effect of screaming?
Matt Standal (voice-over):
Dan Stahler is Yellowstone National Park’s lead wildlife biologist, a job that includes collecting data from sound recorders like this one hidden in a tree near park headquarters.
Dan Stahler:
Another Jeremiah.
Matt Standal (voice-over):
Stahler’s team recorded the barks, yips and howls of Yellowstone’s nine wolf packs, more than 100 wolves, over the past year.
Dan Stahler:
That’s another goal of ours: can we detect unique pack signatures and use them?
Matt Standal (voice-over):
He says: “They collected over 7,000 wolf sounds and were able to identify the acoustic signatures of several wolf packs in the park.
In the future, Stahler believes this bioacoustic work could partly replace the dangerous task of capturing and collaring wolves.
Dan Stahler:
And so what I could imagine a decade from now is we might not have to put collars on certain packs or put collars in certain areas of the park. And then with new cutting-edge AI tools, we hope. We’re not sure yet, but we hope we can answer some really interesting questions about what wolves actually say or can we count wolves? Can we identify unique individuals?
Jeffrey Reed:
Hello everyone. My name is.
Matt Standal (voice-over):
Linguistics researcher and software engineer Dr. Jeff Reed experimented with AI to study wolves near his home, just north of Yellowstone. He puts his technical expertise at the service of the Yellowstone Wolf Project.
Jeffrey Reed:
This is a wolf choir howl and we’re using Google’s AI to see if we can count the number of wolves in a choir howl. So it’s a group of wolves. It’s like you walk into a bar and all the people are talking and you can spot one particular person in the room. Wolves can spot other wolves they know in this cacophony of sound.
Matt Standal (voice-over):
The key to the technology is pattern recognition. According to Reed, these colored patterns are what’s called a spectrogram of wolf howls, representing their strength and frequency over time. Artificial intelligence, he says, can detect patterns and identify individual wolves much faster than any human.
Jeffrey Reed:
These battery-operated devices use AI.
Matt Standal (voice-over):
Reed runs a company that makes high-tech AI-enabled field cameras and audio recorders that Yellowstone uses to monitor its vast space. But these cameras, called Griz cameras, don’t just listen to wildlife. They can also pick up human conversations and activities from hundreds of meters away.
Animal science and human privacy in Yellowstone may soon find themselves on a collision course. 25 of these cameras will be installed in a grid across the park, thanks to a major donation from a company called Colossal Biosciences.
Matt James, Colossal Bioscience:
For me, the question of bioacoustics and wolves is: can we reduce the conflict between wolves and humans?
Matt Standal (voice-over):
Matt James is the Animal Manager at Colossal Biosciences. He claims that AI recording technology can be used to protect wolves from humans.
Matt James:
And can we explain that these are empathetic, emotionally complex animals who are not mindless hunters and who deserve the opportunity to coexist with us.
Matt Standal (voice-over):
Colossal funds $175,000 for Yellowstone bioacoustics study. Additionally, the company hired a team of AI scientists to analyze the data collected by Griz cameras.
Matt James:
We’re really hoping that they can collect tons and tons of data that our team can then start to distill and train the AI to go from just classifying wolf calls to classifying individual calls.
Matt Standal (voice-over):
As we mentioned, all of this data could include the sounds and activities of people in the park. The technology is so new that ethicists are still trying to understand its implications for human privacy in wild places like Yellowstone.
Christopher Preston, University of Montana: This is all the data that can be collected.
Matt Standal (voice-over):
Christopher Preston, a philosophy professor at the University of Montana, studies the ethics of human interactions with the natural world and how technology can shape those interactions.
Christopher Preston:
I mean, if you ask me, would I rather have a wolf thrown out of a helicopter and radio collared or have a wolf listened to by a recording device 24/7? It is clear to me that I prefer the wolf to be listened to through the recording device, as it is a non-invasive technique, much less likely to cause any harm to the animal involved.
Matt Standal (voice-over):
But Preston worries that cameras could inadvertently suck up human sound and images without people knowing they’re being recorded.
Christopher Preston:
We have a different ethic for the human world compared to the wild world. You go into landscapes like that to not be part of a system where people are watching you, where people know what you’re doing and you certainly won’t be able to get away from all of that if it’s possible that your movements are being recorded in a database somewhere.
Matt Standal (voice-over):
As Yellowstone experiments with this controversial new technology, biologist Dan Stahler hopes more Griz Cams will be installed in the coming months. He believes AI-based bioacoustics will help his team better protect these iconic animals as they learn more from each howl and as this ripples across this majestic landscape.
Dan Stahler:
We will continue this study. New questions will emerge. But the fundamental question will be, why is Yellowstone’s wildlife community important to this landscape, important to Montana, and important to the world?
Matt Standal (voice-over):
For PBS News Weekend, I’m Matthew Standal in Yellowstone National Park.