The race to translate animal sounds into human language

In 2025, we will see AI and machine learning harnessed to make real progress in understanding animal communication, answering a question that has puzzled humans for as long as we have existed: “What do the animals each other?” The recent Coller-Dolittle Awardoffering cash prizes of up to half a million dollars for scientists who “crack the code” is an indication of bullish confidence that recent technological developments in machine learning and large language models (LLMs ) put this goal in our grasp.

Many research groups have been working for years on algorithms to make sense of animal sounds. Project Ceti, for example, was decoded the click trains of sperm whales and the songs of humpbacks. These modern machine learning tools require extremely large amounts of data, and until now, such amounts of high-quality, well-annotated data have been lacking.

Consider LLMs like ChatGPT that have training data available to them that includes the entire text available on the Internet. Such information about animal communication has not been accessible in the past. It is not only that the human data corpus is many orders of magnitude larger than the type of data we have access to for animals in nature: more than 500 GB of words were used to form GPT-3, compared to little more than 8,000 “tails”. ” (or vocalizations) for the recent Ceti Project analysis of sperm whale communication.

Also, when working with human language, we already have to know what is said. We also know what constitutes a “word,” which is a huge advantage over interpreting animal communication, where scientists rarely know whether a particular wolf howl, for example, means something different from another wolf howl. , or even if wolves consider a howl as. somehow analogous to a “word” in human language.

However, 2025 will bring new advances, both in the amount of animal communication data available to scientists, and in the types and power of AI algorithms that can be applied to that data. Automatic recording of animal sounds has been put within easy reach of any scientific research group, with low-cost recording devices such as the AudioMoth exploding in popularity.

Massive datasets are now online, as recorders can be left in the field, listening to the calls of gibbons in the jungle or birds in the forest, 24/7, over long periods of time. There were occasions when such massive datasets were impossible to manage manually. Now, new automatic detection algorithms based on convolutional neural networks can run through thousands of hours of recordings, picking out animal sounds and grouping them into different types, based on their natural acoustic characteristics.

Once those large datasets of animals are available, new analytical algorithms become a possibility, such as the use of deep neural networks to find a hidden structure in sequences of animal vocalizations, which can be analogous to the meaningful structure in language human

However, the fundamental question that remains unclear is, what exactly are we hoping to do with these animal sounds? Some organizations, such as Interspecies.io, state their goal quite clearly as, “to transduce signals from one species into coherent signals for another.” In other words, to translate animal communication in human language. However, most scientists agree that non-human animals do not have a language of their own, at least not in the way that we humans have language.

The Coller Dolittle Prize is a little more sophisticated, looking for a way “to communicate with or decipher the communication of an organism.” Deciphering is a slightly less ambitious goal than translation, considering the possibility that animals do not, in fact, have a language that can be translated. Today we do not know how much information, or how little, animals transmit between themselves. By 2025, humanity will have the potential to leapfrog our understanding of not only how much animals say, but also what exactly they say.