Artificial intelligence (AI) is on the cusp of making significant strides in understanding animal communication, a long-standing puzzle for humans. The recent Coller-Dolittle Prize offers a $500,000 cash prize for scientists who crack the code, reflecting growing confidence in AI’s potential to decipher animal sounds.
Several research groups have been working on algorithms to make sense of animal noises, such as Project Ceti, which decodes whale clicks and songs. However, traditional machine learning tools required vast amounts of high-quality data, a challenge that recent large language models (LLMs) like ChatGPT address. LLMs have access to massive amounts of human text data, including the internet’s entirety.
Unlike animal communication, where context and meaning are unclear, humans already possess a framework for understanding words and language. AI algorithms can now process vast datasets, automatically detecting and clustering animal sounds into different types based on acoustic characteristics. This technology enables scientists to analyze sequences of vocalizations and uncover hidden structure, similar to human language patterns.
However, the ultimate goal remains uncertain: what do animals communicate with each other? While some organizations aim to translate animal signals into human language, most scientists agree that non-human animals lack a distinct language. The Coller-Dolittle Prize seeks to decipher an organism’s communication, acknowledging that not all animals possess a language that can be translated.
In 2025, humanity may leapfrog its understanding of animal communication, uncovering not only how much information animals convey but also what they are saying to each other. As AI advances, the potential for breakthroughs in animal communication grows, promising new insights into the complex world of wildlife interactions.
Source: https://www.wired.com/story/artificial-intelligence-translation-animal-sounds-human-language