AI and drones are decoding whale calls. What they're finding could change how we think about animal culture.

Key points

• Researchers used hydrophones, underwater microphones, to record beluga whale calls in the Canadian Arctic as far back as 2015.
• Belugas produce dozens of distinct sounds, which scientists believe help them stay connected in dark, murky Arctic water.
• AI tools are now helping researchers sort and analyse thousands of hours of whale audio far faster than any human team could.
• Other whale species, including humpbacks and sperm whales, also show signs of learned vocal traditions passed between individuals.
• Scientists say understanding animal communication could reshape what we know about culture and language outside of humans.

In 2015, whale researcher Valeria Vergara pitched a small tent at the edge of an icy bay on Somerset Island, deep in the Canadian Arctic. Fog was so thick she could barely see her hands. She dropped a hydrophone, an underwater microphone, into the freezing water and waited.

What she recorded was extraordinary.

Up to 1,000 beluga whales, many travelling with newborn calves, filled the surrounding bay with near-constant noise. Clicks, chirps, whistles, squeals. Dozens of different sounds, all from animals whose survival may depend on staying in contact through dark, cloudy Arctic water.

Why does any of this matter to the rest of us?

This is where AI enters the picture, and where a story about remote Arctic fieldwork becomes something with broader stakes.

Manually listening to thousands of hours of underwater recordings is slow, expensive, and prone to human error. Researchers now feed those recordings into machine learning systems, a type of AI that finds patterns in large amounts of data, to sort and label calls far faster than any team of human listeners could manage. What once took years can now take weeks.

The Guardian first reported on this wave of AI-assisted wildlife research, which is uncovering evidence that whales do not just communicate. They may actually teach each other.

Humpback whales, for instance, appear to pick up new songs from neighbouring groups, spreading specific call patterns across ocean populations the way a catchy tune spreads through a crowd. Sperm whales use distinctive rhythmic clicking patterns, known as codas, that differ between social groups called clans, and young sperm whales seem to learn their clan's coda rather than being born knowing it.

That last point matters. Learned behaviour passed through a social group is one of the defining features of culture. If whales do it, the line between human culture and animal behaviour gets a lot blurrier.

What does this mean for whale conservation?

Decoding communication could directly help protect species at risk. Beluga populations face pressure from climate change, shipping noise, and habitat loss. If scientists can identify specific calls linked to feeding, stress, or mother-calf bonding, wildlife managers gain a practical tool: a way to detect distress or disruption in real time, using acoustic monitoring rather than expensive boat surveys.

Drones add another layer, letting researchers film whale behaviour above water at the same moment hydrophones capture what is happening below, matching sounds to actions in a way that was nearly impossible a decade ago.

Common questions

Do whales actually have a language?

Not in the way humans do. Whale communication is complex and learned, but scientists have not yet established that it carries the kind of grammar or open-ended meaning that defines human language. The research suggests it is far richer than simple instinct, though.

How does the AI actually help researchers?

The AI acts like a very fast librarian. It listens to recordings, groups similar sounds together, and flags patterns, freeing biologists to focus on what those patterns mean rather than spending years just cataloguing raw audio.

Can this research really help save whales?

Potentially, yes. Identifying stress calls or disrupted communication patterns could give conservationists an early warning system for threats like shipping traffic or habitat change, making protection efforts faster and more targeted.

The honest takeaway: AI is not replacing the scientist in the tent at the water's edge. It is giving her a far better chance of understanding what she hears.