Neural networks are capable of translating between languages--in some cases even between two languages where there is little or no access to parallel translations, in what is known as Unsupervised Machine Translation (UMT). Given this progress, it is intriguing to ask whether machine learning tools can ultimately enable understanding animal communication, particularly that of highly intelligentanimals. We propose a theoretical framework for analyzing UMT when no parallel translations are available and when it cannot be assumed that the source and target corpora address related subject domains or posses similar linguistic structure. Weexemplify this theory with two stylized models of language, for which our framework provides bounds on necessary sample complexity; the bounds are formally proven and experimentally verified on synthetic data. These bounds show that the error rates are inversely related to the language complexity and amount of common ground. This suggests that unsupervised translation of animal communication may be feasible if the communication system is sufficiently complex.

Current bioacoustic AI systems achieve impressive cross-species performance by processing animal communication through transformer architectures, foundation model paradigms, and other computational approaches. However, these approaches overlook a fundamental question: what happens when one form of recursive cognition--AI systems with their attention mechanisms, iterative processing, and feedback loops--encounters the recursive communicative processes of other species? Drawing on philosopher Y uk Hui's work on recursivity and contingency, I argue that AI systems are not neutral pattern detectors but recursive cognitive agents whose own information processing may systematically obscure or distort other species' communicative structures. This creates a double contingency problem: each species' communication emerges through contingent ecological and evolutionary conditions, while AI systems process these signals through their own contingent architectural and training conditions. I propose that addressing this challenge requires reconceptualizing bioacoustic AI from universal pattern recognition toward diplomatic encounter between different forms of recursive cognition, with implications for model design, evaluation frameworks, and research methodologies.

If you had an AI Whale-to-English translator, how could you validate whether or not it is working? Does one need to interact with the animals or rely on grounded observations such as temperature? We provide theoretical and proof-of-concept experimental evidence suggesting that interaction and even observations may not be necessary for sufficiently complex languages. One may be able to evaluate translators solely by their English outputs, offering potential advantages in terms of safety, ethics, and cost. This is an instance of machine translation quality evaluation (MTQE) without any reference translations available. A key challenge is identifying ``hallucinations,'' false translations which may appear fluent and plausible. We propose using segment-by-segment translation together with the classic NLP shuffle test to evaluate translators. The idea is to translate animal communication, turn by turn, and evaluate how often the resulting translations make more sense in order than permuted. Proof-of-concept experiments on data-scarce human languages and constructed languages demonstrate the potential utility of this evaluation methodology. These human-language experiments serve solely to validate our reference-free metric under data scarcity. It is found to correlate highly with a standard evaluation based on reference translations, which are available in our experiments. We also perform a theoretical analysis suggesting that interaction may not be necessary nor efficient in the early stages of learning to translate.

Abstract symbolic writing systems are semiotic codes that are ubiquitous in modern society but are otherwise absent in the animal kingdom. Anthropological evidence suggests that the earliest forms of some writing systems originally consisted of iconic pictographs, which signify their referent via visual resemblance. While previous studies have examined the emergence and, separately, the evolution of pictographic writing systems through a computational lens, most employ non-naturalistic methodologies that make it difficult to draw clear analogies to human and animal cognition. We develop a multi-agent reinforcement learning testbed for emergent communication called a Signification Game, and formulate a model of inferential communication that enables agents to leverage visual theory of mind to communicate actions using pictographs. Our model, which is situated within a broader formalism for animal communication, sheds light on the cognitive and cultural processes that led to the development of early writing systems.

Neural networks are capable of translating between languages--in some cases even between two languages where there is little or no access to parallel translations, in what is known as Unsupervised Machine Translation (UMT). Given this progress, it is intriguing to ask whether machine learning tools can ultimately enable understanding animal communication, particularly that of highly intelligentanimals. We propose a theoretical framework for analyzing UMT when no parallel translations are available and when it cannot be assumed that the source and target corpora address related subject domains or posses similar linguistic structure. Weexemplify this theory with two stylized models of language, for which our framework provides bounds on necessary sample complexity; the bounds are formally proven and experimentally verified on synthetic data. These bounds show that the error rates are inversely related to the language complexity and amount of common ground.

In 2025 we will see AI and machine learning leveraged to make real progress in understanding animal communication, answering a question that has puzzled humans as long as we have existed: "What are animals saying to each other?" The recent Coller-Dolittle Prize, offering cash prizes up to half-a-million dollars for scientists who "crack the code" is an indication of a bullish confidence that recent technological developments in machine learning and large language models (LLMs) are placing this goal within our grasp. Many research groups have been working for years on algorithms to make sense of animal sounds. Project Ceti, for example, has been decoding the click trains of sperm whales and the songs of humpbacks. These modern machine learning tools require extremely large amounts of data, and up until now, such quantities of high-quality and well-annotated data have been lacking. Consider LLMs such as ChatGPT that have training data available to them that includes the entirety of text available on the internet.

You see a cat chasing a mouse. You probably don't realize it, but as soon as you catch sight of this scene unfolding, your brain makes a key distinction between the cat and the mouse: It identifies who's chasing, and who's being chased. This capacity to distinguish between the "agent" (the entity performing an action) and the "patient" (the entity upon which that action is being performed) is called "event decomposition," and it's long been thought that it was unique to humans. However, a new study published in PLOS Biology on November 26 suggests that this is not the case: great apes (specifically gorillas, chimpanzees, and orangutans) also seem to track events in the way that we do, distinguishing between agent and patient. This finding is notable because scientists believe event decomposition lies at the heart of something that is unique to humans.

Similar to humans, animals make extensive use of verbal and non-verbal forms of communication, including a large range of audio signals. In this paper, we address dog vocalizations and explore the use of self-supervised speech representation models pre-trained on human speech to address dog bark classification tasks that find parallels in human-centered tasks in speech recognition. We specifically address four tasks: dog recognition, breed identification, gender classification, and context grounding. We show that using speech embedding representations significantly improves over simpler classification baselines. Further, we also find that models pre-trained on large human speech acoustics can provide additional performance boosts on several tasks.

Being able to chat with your dog or finally understand what your cat wants would be a dream come true for many pet owners. But this scientist, dubbed'the real life Doctor Dolittle', says he already can. Dr Arik Kershenbaum, an expert in animal communications from the University of Cambridge, says that everything from a dog's whine to a dolphin's whistle is packed with meaning. But, has Dr Kershenbaum really cracked the animal code or is he barking up the wrong tree? MailOnline spoke with the real-life animal whisperer to put his skills to the test.

If an unexpected meow, peculiar pose, or unusual twitch of the whiskers leaves you puzzling over what your cat is trying to tell you, artificial intelligence may soon be able to translate. Scientists are turning to new technology to unpick the meanings behind the vocal and physical cues of a host of animals. "We could use AI to teach us a lot about what animals are trying to say to us," said Daniel Mills, a professor of veterinary behavioural medicine at the University of Lincoln. Previous work, including by Mills, has shown that cats produce a variety of facial expressions when interacting with humans, and this week researchers revealed felines have a range of 276 facial expressions when interacting with other cats. "However, the facial expressions they produce towards humans look different from those produced towards cats," said Dr Brittany Florkiewicz, an assistant professor of psychology at Lyon College in Arkansas who co-authored the new work.