AI is helping scientists decode birdsong — and human-animal conversations may be closer than ever

Dr Julie Elie of the University of California, Berkeley, received the 2026 Coller-Dolittle Prize for Two-Way Interspecies Communication after her research showed that zebra finches use 11 distinct call types linked to specific situations. By combining years of field observations with machine-learning analysis and behavioural experiments, her team found the birds responded to calls according to their meaning rather than simply their sound.

The findings, published in The Guardian’s coverage of the award, suggest zebra finches recognise both individual callers and the message being conveyed — a level of communication that is more structured than previously understood.

The project was anything but a quick AI success story.

Elie’s team spent more than a decade recording thousands of vocalisations before using machine learning to organise and analyse the data. The technology accelerated the process, but only after years of painstaking observation had established the context behind each call.

Behavioural tests produced one of the strongest pieces of evidence. When researchers played back different calls, the birds were more likely to mistake calls carrying similar meanings than those that merely sounded alike, suggesting they were responding to information rather than acoustic patterns.

Researchers have recently reported evidence that bonobos combine calls into sequences resembling simple linguistic rules, while separate studies have documented sophisticated vocal systems in chimpanzees and ultrasonic communication among African striped mice. Each discovery adds another piece to a puzzle that scientists are only beginning to assemble.

The idea of talking to animals still belongs more to science fiction than everyday science.

Even with increasingly powerful AI models, researchers first have to establish whether a sound carries a consistent meaning, reflects an emotional state or simply signals an instinctive response. Context matters, and that remains one of the biggest challenges.

Even so, the pace of progress has accelerated sharply over the past few years. With AI processing datasets that would once have taken decades to analyse, scientists say understanding animal communication no longer feels like an impossible ambition — only a difficult one.