Scientists make stunning AI-powered discoveries while listening to animal recordings: 'It worked extremely well'

Researchers from James Cook University have used advanced artificial intelligence applications to analyze more than 300,000 hours (roughly 36 years) of mammal calls across eastern Australia.

It's hoped that the data will significantly improve wildlife monitoring and conservation efforts in the region. 

A research summary from the university explained the data covered species spanning from Far North Queensland to southern New South Wales. 

The team compared various methods of tracking mammal population health, including passive acoustic monitoring, camera trapping, and direct observation. 

They noted that while the first technique has worked well for vocalizing birds, it hasn't been tested much on mammals, which inspired their study, published in the journal Methods in Ecology and Evolution. 

Dr. Sebastian Hoefer, lead researcher and JCU Postdoctoral Research Fellow, explained that Australia has the worst mammal extinction rate in the world, and most are found nowhere else. They play a critical role in ecological health and biodiversity, but human activities have severely limited their range, leaving them vulnerable to environmental changes. 

That's why monitoring their movements and population health is crucial for conservation, but large-scale observation isn't easy. Thankfully, the open-source BirdNET deep learning model has erased many of the difficulties, allowing the team to access vast amounts of data with far less effort. 

By inputting a mammal call into the system, such as a male koala's bellow, the AI platform can detect similar sounds across all datasets.

"It worked extremely well, especially for long-term monitoring," Dr. Hoefer said.

"Now we can rely on acoustic monitoring and AI for vocal mammal species, and this allows us to redirect more time and funding towards monitoring mammals that don't vocalize."

Should we be creating robots that act like living beings? No way 🙅 Only for certain uses 🤖 I'm not sure 🤷 Totally 💯 Click your choice to see results and speak your mind.

Using the BirdNET system along with remote sensing techniques can give the team new insights into when certain species are most active, helping them "plan targeted management or monitoring during those times."

Some environmental advocates have warned about the potential negative impacts of AI, including massive energy and water usage, resource depletion from hardware manufacturing, and higher energy bills for consumers. 

However, AI has proved indispensable in wildlife conservation, as it can analyze large datasets to monitor species, detect poachers, forecast habitat loss, and inform land-use planning.  

For example, conservation teams in China are using AI to monitor critically endangered Yangtze finless porpoises. Thanks to these and other restoration efforts, the species' population has increased by over 23% in recent years, marking the first rise in decades. 

In the United States, a drone company and an AI firm based in the Upper Midwest teamed up to create POLLi, a conservation software platform to identify milkweed for monarch butterfly conservation. 

Monarchs are crucial pollinators and support ecosystem health in many regions. Despite surprising rebounds in both Western and Eastern populations, the U.S. Fish and Wildlife Service proposed a federal threatened listing under the Endangered Species Act in late 2024, underscoring the need for ongoing conservation efforts. 

The latest use of AI to track mammal calls could have important implications for global research and species recovery. In the future, researchers aim to create a "species-specific recogniser" to track the mammals they recorded during the study.

"This will make it even easier to detect and monitor mammal species across Australia in the future," Dr. Hoefer said.

Get TCD's free newsletters for easy tips to save more, waste less, and make smarter choices — and earn up to $5,000 toward clean upgrades in TCD's exclusive Rewards Club.