Google has introduced a unique AI model, DolphinGemma, designed to understand and generate dolphin sounds. The model was developed together with researchers from the Wild Dolphin Project and the University of Georgia, and it was trained on a 40-year archive of acoustic observations of Atlantic spotted dolphins.
At the core of DolphinGemma is a language model with 400 million parameters that works using modern AI principles, but applied to dolphin clicks and whistles. It not only analyzes the animals’ “speech,” but can also reproduce responses in real time.
The practical component is supported by the CHAT system—a dive computer built into modified Google Pixel smartphones. Using synthetic sounds and tracking dolphins’ reactions, researchers build a basis for communication between species.
This is the first AI system in the world that can operate outside the lab and on mobile devices. It is not only a technological breakthrough, but a foundation for a new level of ethical interaction with other intelligent beings.
And what would you like to talk about with dolphins?
