How AI is decoding whale communication : Short Wave : NPR
wildestanimal/Getty Images
wildestanimal/Getty Images
Scientists are testing the limits of artificial intelligence when it comes to language learning.
One recent challenge? Learning whale.
Researchers are using machine learning to analyze and decode whale sounds — and it’s just as complicated as it seems.
Sperm whales are interesting to researchers because they have big brains, close family groups and coordinate in lots of ways. They dive and hunt together, and sperm whale biologist Shane Gero says they even babysit for each other.
“It’s hard not to see cousins playing while chatting,” he says. “To not see moms hand over to a babysitter and exchange a few words before walking out the door, so to speak, to go eat in the deep ocean.”
Sperm whales also spend a lot of time in the dark. They dive thousands of feet down, searching for deep-sea squid, so Gero says sound is everything to them. He studies these whales in the Caribbean with the Dominica Sperm Whale Project. And he and other researchers have heard the exchanges of these whales for years — they communicate with clicks, often talking over one another or forming different arrangements and patterns.
But sperm whale scientists like Gero have long wondered what these clicks mean. So they teamed up with AI researchers in a collaboration called Project CETI to decode the sounds they’ve studied for years. So far, the researchers have identified what they’re calling a sperm whale phonetic alphabet.
Curious about other mysteries of nature? Email us at shortwave@npr.org.
Listen to Short Wave on Spotify, Apple Podcasts and Google Podcasts.
Listen to every episode of Short Wave sponsor-free and support our work at NPR by signing up for Short Wave+ at plus.npr.org/shortwave.
Today’s episode was produced by Rachel Carlson. It was edited by Rebecca Ramirez. Lauren Sommer checked the facts. Stacey Abbott was the audio engineer.
