| AI and ML Used To Understand Bird Calls |
| Written by Sue Gee | ||| | Sunday, 19 July 2026 | ||| | The Coller Dolittle Challenge for Two-Way Interspecies Communication has awarded its annual $100,000 prize to Dr Julie Elie, a scientist at the University of California, Berkeley, who with the help of machine learnming and AI has contructed an 11-word dictionary of "words" used by Zebra Finches. This prize, inspired by the Turing Test, is awarded for the most promising research aimed at developing an algorithm for communicating with animals. The challenge is under the auspices of the Jeremy Coller Foundation, a UK-based philanthropic organization in partnership with Tel Aviv University in Israel and its name combines that of the foundation's creator, Jeremy Coller, with the fictional character Dr. Dolittle, who could talk to animals. Launched in 2024, the initiative features an annual progressive prize of $100,000 and larger Grand Prize, either a $10 million equity investment or $500,000 in cash, for the team that successfully achieves true, independent, two-way communication where the animal interacts without recognizing it is dealing with humans. To win the annual award, submissions must meet strict scientific criteria. The research must decipher or interface with animal communication without being invasive, it must demonstrate understanding or interaction across more than one context (such as foraging, mating, or alarm calls) using the organism's own natural signals, and must prove a measurable reaction from the organism when these signals are broadcast back to it. In 2025, the inaugural prize was awarded to a research team led by Dr Laela Sayigh from the Woods Hole Oceanographic Institution for using AI to analyze and decipher the signature whistle types and communication patterns of wild bottlenose dolphins.. This year it was won by Dr Julie Yee from the University of California, Berkeley for her 15-year study mapping out an 11-word core "dictionary" of zebra finch vocalizations and demonstrating that the birds group their calls based on behavioral meaning rather than just acoustic similarity. Zebra finches are constant vocalizers, resulting in thousands of hours of dense audio data. Traditional manual audio analysis would take years to sort through this volume but AI algorithms parsed these massive datasets at high speeds, systematically detecting, isolating, and tracking individual acoustic events across varying contexts. Machine learning models were employed to extract and categorize the precise acoustic features of the finch vocalizations and AI tools uncovered complex bioacoustic patterns, including subtler variations that are difficult for human hearing to differentiate, to map out the structural differences between the 11 call-types that make up the zebra finch "dictionary". While artificial intelligence and machine learning were fundamental to creating the dictionary, it was classic behavorism that was used to test the results and to confirm that the sounds had the meanings ascribed with an experiment in which birds were trained to interact with playback systems. By mapping the precise acoustic distances calculated by the AI against the birds' behavioral responses, the researchers noticed a critical anomaly: the birds' classification errors did not align with acoustic similarity. The birds regularly confused acoustically distinct calls that shared a similar behavioral meaning (like long-distance and short-distance contact calls), while easily distinguishing highly similar sounds that meant completely different things, leading to the conclusion that the birds group their calls based on behavioral meaning rather than just acoustic similarity. In this TV news interview, Julie Elie explains the experiment she devised and goes on to explain how being able to talk to birds might be beneficial to them:
More Information #
Related Articles #
To be informed about new articles on I Programmer, sign up for our
Comments #
or email your comment to: | ||| | Last Updated ( Sunday, 19 July 2026 ) |