Scientists are using artificial intelligence to decode whale communication and translate it into human language.

Project CETI (Cetacean Translation Initiative), a non-profit launched in 2020, applies natural language processing to decode sperm whale communication. Sperm whales possess the largest brains on Earth and live in matrilineal societies with distinct cultural groups. They communicate through codas—rhythmic patterns of clicks that researchers believe carry sophisticated information about their social structures and environment.The project uses underwater hydrophones to record millions of whale clicks near Dominica. These massive datasets are processed with deep learning algorithms similar to those powering GPT-4 to identify recurring syntax and grammatical patterns. Leading researchers including David Gruber from the City University of New York and Robert Wood from Harvard University combine specialized robotics and AI to map the language of the deep sea.One major challenge is the speed and frequency of whale clicks, which occur in bursts lasting only milliseconds. Human hearing cannot distinguish the subtle variations in timing and frequency that carry meaning. AI systems, however, can detect these nuances with precision. By 2024, the team had identified over 150 distinct coda types used by different sperm whale clans. The ultimate goal is to develop a chatbot-like interface that can generate whale-like responses, allowing researchers to test whether they can accurately interpret whale intent and establish genuine two-way communication.