AI Unlocks Animals' Emotional Languages
Emily Ma '28
For decades, researchers have attempted to understand animal emotions to enhance their welfare and management by detecting physical and vocal signs. However, human observations often proved inaccurate and subjective, with many discoveries being limited to only one species. Recently, researchers at the University of Copenhagen utilized artificial intelligence (AI) to open new paths for interpreting animal emotions in a standardized and efficient way.
Bioacoustics, a field focused on analyzing animal sounds, emerged in the mid-1900s as a systematic approach to understand animal emotions. In a study, University of Copenhagen researcher E. F. Briefer proved the reliability of the approach, discovering that “vocal indicators of emotions in animals could represent convenient and non-invasive indicators” (Briefer, 2012). The ability to detect emotions such as happiness, fear, or irritation using “changes in duration, energy distribution, fundamental frequency, and amplitude modulation” has created opportunities for AI to take on this interpretive task (University of Copenhagen, 2025). Because AI is able to process and analyze massive datasets, detect sound frequencies that humans cannot hear, and classify information with remarkable precision, the incorporation of AI could produce huge strides in animal emotion research and improvement in animal welfare.
Researchers at the University of Copenhagen created and trained a machine learning model on thousands of animal vocalizations to allow it to identify subtle markers of emotion in animals. Small differences in call duration, energy distribution, fundamental frequency, and amplitude modulation were used to teach the model how to differentiate between positive and negative emotions. The model achieved a shocking accuracy rate of 89.49% across seven ungulate (hoof mammal) species (University of Copenhagen, 2025). This breakthrough is one of the few successful recorded machine learning models where AI has been fused with the study of animal emotions and the “first cross-species study to detect emotional valence using AI” (University of Copenhagen, 2025).
Similarly, in another research conducted by the Cetacean Translation Initiative (CETI), researchers constructed a Natural Language Processing (NLP) algorithm that interprets the clicks and whistles of dolphins (Hashemi, 2025). The algorithm enabled researchers to “listen to the dolphins’ vocalizations, identify patterns and predict what comes next, just as LLMs do with human language” (Hashemi, 2025).
With this new discovery in the field of animal emotions, the potential applications of AI in agriculture, animal rights, and animal welfare is far-ranging. In the agricultural industry, AI systems could help monitor farm animals and alert their caretakers when the systems detect signs of sickness, stress, or pain in the animal long before their illnesses become visible. This advancement leads to more humane and efficient farming practices by reducing the spread of disease and improving productivity. Beyond agriculture, AI-driven emotion recognition can be implemented in zoos, wildlife conservation, veterinary care, and animal shelters. By allowing workers to understand how animals are feeling more precisely and efficiently, AI can improve ethical standards and encourage empathetic treatment towards animals. In addition, the findings of the study support the idea that emotional states are not uniquely human, suggesting that animals also experience a spectrum of feelings that can now be objectively measured and understood with the help of AI. The proof of their emotions lends weight to arguments for animal rights and welfare as it consolidates the existence of animal sentience and consciousness.
Despite the model’s potential to improve welfare, it can also be used to optimize productivity in factory farming without any real concern for the animal’s well-being. This technology marks a meaningful step towards bridging the communication gap between humans and animals, but ethical considerations would also need to be taken into account in order to assure that this technology is well-used for the intended purposes.
References
University of Copenhagen - Faculty of Science. (2025, February 21). AI unlocks the emotional language of animals. ScienceDaily. Retrieved from https://www.sciencedaily.com/releases/2025/02/250221125552.htm
Briefer, E.F. (2012), Vocal communication of emotions. J Zool, 288: 1-20. Retrieved from https://doi.org/10.1111/j.1469-7998.2012.00920.x
Hashemi, S. (2025, April 25). Google is training a new A.I. model to decode dolphin chatter—and potentially talk back. Smithsonian Magazine. https://www.smithsonianmag.com/smart-news/google-is-training-a-new-ai-model-to-deco de-dolphin-chatter-and-potentially-talk-back-180986434/