Researchers have long tried to understand how animals communicate. What’s different today is the incredible advances in deep learning and artificial intelligence, said Bronstein, a professor of machine learning and pattern recognition at Imperial College London. He cited the language-generation tool GPT-3, which can produce human text on demand — answer questions, write poems, stories, essays, or news reports — as an example of groundbreaking technology that can process massive amounts of data, specifically words, with speed and accuracy. Their work will rely on a similar model.

To produce a “whale grammar,” the researchers will first collect sounds using specially designed computers attached to the animals with suction cups, underwater microphones, and swimming robots. They’ll then use traditional processors on land for the heavy computational lifting. “What is important is that we are building all these data sets for automated analysis by machine-learning algorithms rather than by humans, by hand, as it was done in the past,” said Bronstein. “And we currently estimate that it will be able to collect billions of sperm whale clicks and associated metadata that will result in a data set that is similar in order of magnitude to those used portraying human language models.”

Far from the murderous beast depicted by Melville, sperm whales are “truly gentle creatures,” said Gruber. They live in matrilineal pods in which the mothers and grandmothers dictate interactions. Those complex social lives, along with their large brains and reliance on sound to communicate at great depths — they often dive a mile below the surface and can hold their breath for more than an hour — make them ideal study subjects.

For Goldwasser, a theorist, developing a robust set of questions is crucial to the work. Through CETI, she is hoping to understand the goal of whale communication, and whether it extends beyond the search for food and efforts to simply locate one another while hunting in the deep. She is also fascinated by the ways humans “develop very complicated error correction methods for phones or computers, because there’s noise on the line,” and whether whales use similar corrections when communicating over long distances in water. Getting a better handle on the complexity of whale communication might also help scientists better understand other non-human communication and whether there is a “hierarchy of complexity” in animal communication, one possibly “connected to evolution,” said Goldwasser, professor in electrical engineering and computer science at the University of California, Berkeley.

Well aware that other researchers have long been involved in studying whale and dolphin communication, Gruber said they welcome collaboration with other scientists and with the public at large. “We’re basically putting everything up as open source, and we’re going to start a citizen science community that could help come on the ride with us.”

Acknowledging that they are just at “the beginning of this journey,” Bronstein said he hopes their work “leads to some shift in the way we treat our environment. And maybe it will result in more respect for the living world.”