AI Gliders Help Scientists Decode Sperm Whale Communication

Researchers from Project CETI are using AI-enabled underwater gliders to track sperm whales and decode their click-based communication alphabet.

This technology overcomes a significant barrier in marine biology because sperm whales dive deeper than one kilometer ^[1]. Traditional tagging methods often fail at these depths and during the long periods the animals remain submerged, which has historically limited the ability of scientists to study their complex social interactions.

The gliders utilize buoyancy-powered propulsion to navigate the deep waters off Germany and surrounding North Atlantic regions ^[2]. By following the whales, these drones can record the acoustic signals the animals use to communicate. These signals, known as codas, are sequences of clicks that researchers believe function as a structured alphabet ^[2].

Project CETI has been tracking the whales using these gliders for several months ^[3]. The integration of artificial intelligence allows the gliders to maintain a presence near the whales, capturing high-quality data that was previously inaccessible. This data is then processed to identify patterns in the clicks, helping the team map out the phonetic structure of the whales' language ^[2].

The project focuses on the North Atlantic, where the deep-sea environment provides a natural laboratory for observing these mammals. The buoyancy-driven design of the gliders allows them to operate for extended periods without the need for heavy batteries or loud engines, minimizing the impact on the animals' behavior ^[2].

By combining autonomous robotics with machine learning, the team aims to translate the meaning behind the clicks. The goal is to understand not just the sounds, but the intent and information being shared between pods in the deep ocean ^[2].