Skip to main content
Energy

Google unveils AI model to understand dolphin communication

Listen
Share

  • Google has announced DolphinGemma, a new AI model developed in partnership with Georgia Tech and the Wild Dolphin Project. The project aims to help humans better understand dolphin communication.
  • DolphinGemma uses decades of dolphin vocalization data to identify relationships between dolphin sounds and social interactions. It has the potential to predict future vocal sequences.
  • Google plans to release DolphinGemma as an open model sometime in 2025. The goal is to enable broader research and potentially create a shared human-dolphin communication system.

Full Story

On National Dolphin Day, Google announced the development of a new artificial intelligence model aimed at deciphering dolphin communication. Dubbed DolphinGemma, it was developed in collaboration with researchers from Georgia Tech and the Wild Dolphin Project.

Media Landscape

See how news outlets across the political spectrum are covering this story. Learn more
Left 38% Center 50% Right 11%
Bias Distribution Powered by Ground News

“This approach in the quest for interspecies communication pushes the boundaries of AI and our potential connection with the marine world,” Google said in a statement.

QR code for SAN app download

Download the SAN app today to stay up-to-date with Unbiased. Straight Facts™.

Point phone camera here

How can AI help humans understand dolphins?

For about 40 years, the Wild Dolphin Project has studied a population of wild Atlantic spotted dolphins in the Bahamas. Their research focuses on linking specific dolphin vocalizations with behaviors such as mother-calf calls and mating displays. They are now using this decades-worth of data to train DolphinGemma.

How does DolphinGemma work?

Built on the same foundational technology as Google’s Gemini AI models, DolphinGemma is designed to analyze and generate dolphin-like sound sequences. According to Google, the model can process complex vocal patterns to identify relationships between sounds and social interactions. It has the potential to predict future vocal sequences.

Developers optimized DolphinGemma to run on Google Pixel smartphones, enabling real-time field analysis without the need for specialized equipment.

What happens next?

Researchers involved in the project said the technology could enhance their ability to interpret dolphin communication more quickly and accurately. Google also announced plans to release DolphinGemma as an open model later this year. It would allow other researchers to adapt it for studying additional dolphin populations.

“By identifying recurring sound patterns, clusters and reliable sequences, the model can help researchers uncover hidden structures and potential meanings within the dolphins’ natural communication,” Google said. “Eventually, these patterns, augmented with synthetic sounds created by the researchers to refer to objects with which the dolphins like to play, may establish a shared vocabulary with the dolphins for interactive communication.”

Tags: ,

[Jack Aylmer]

TODAY IS NATIONAL DOLPHIN DAY-

AND GOOGLE IS USING THIS AS A CHANCE TO ANNOUNCE A NEW DEVELOPMENT IN ARTIFICIAL INTELLIGENCE AIMED AT BETTER UNDERSTANDING DOLPHIN COMMUNICATION.

IN COLLABORATION WITH RESEARCHERS FROM GEORGIA TECH AND THE WILD DOLPHIN PROJECT, THE TECH GIANT UNVEILED DOLPHINGEMMA-

AN AI MODEL DESIGNED TO ANALYZE AND GENERATE DOLPHIN-LIKE SOUND SEQUENCES.

FOR NEARLY FOUR DECADES, THE WILD DOLPHIN PROJECT HAS STUDIED A SPECIFIC POPULATION OF WILD ATLANTIC SPOTTED DOLPHINS IN THE BAHAMAS-

EFFORTS WHICH AIM TO LINK DOLPHIN VOCALIZATIONS WITH SPECIFIC ACTIONS AND SOCIAL INTERACTIONS-

FROM MOTHER-CALF CALLS TO MATING DISPLAYS.

NOW, ALL THOSE YEARS OF DATA ARE BEING USED TO TRAIN DOLPHINGEMMA.

BUILT ON THE SAME TECHNOLOGY AS GOOGLE’S GEMINI FAMILY OF AI MODELS-

IT PROCESSES COMPLEX DOLPHIN VOCAL SEQUENCES TO IDENTIFY PATTERNS AND POTENTIALLY PREDICT WHAT A DOLPHIN MIGHT SAY NEXT.

 

THE MODEL RUNS DIRECTLY ON GOOGLE PIXEL SMARTPHONES USED IN THE FIELD-

ALLOWING FOR REAL-TIME SOUND ANALYSIS AND RESPONSE WITHOUT THE NEED FOR BULKY CUSTOM EQUIPMENT.

RESEARCHERS SAY THIS COULD ENHANCE THEIR ABILITY TO INTERPRET DOLPHIN COMMUNICATION MORE QUICKLY AND ACCURATELY.

GOOGLE PLANS TO MAKE DOLPHINGEMMA AN OPEN MODEL LATER THIS YEAR-

SO OTHER RESEARCHERS CAN ADAPT IT FOR STUDYING DIFFERENT DOLPHIN SPECIES.

 

ALTHOUGH IT’S STILL EARLY DAYS, THIS COMBINED EFFORT OF FIELD BIOLOGISTS, ENGINEERS, AND AI DEVELOPERS HOPE TO BRING HUMANS CLOSER TO UNDERSTANDING ONE OF NATURE’S MOST INTELLIGENT MARINE MAMMALS.

 

FOR STRAIGHT ARROW NEWS, I’M JACK AYLMER.