Arquivo da tag: Acoplamento humano-animal

AI Is Deciphering Animal Speech. Should We Try to Talk Back? (Gizmodo)

gizmodo.comoriginal article

As scientists use machine learning to decode the sounds of whales, dogs, and dolphins, opinions vary on how best to deploy the technology.

Isaac Schultz

May 17, 2025


Chirps, trills, growls, howls, squawks. Animals converse in all kinds of ways, yet humankind has only scratched the surface of how they communicate with each other and the rest of the living world. Our species has trained some animals—and if you ask cats, animals have trained us, too—but we’ve yet to truly crack the code on interspecies communication.

Increasingly, animal researchers are deploying artificial intelligence to accelerate our investigations of animal communication—both within species and between branches on the tree of life. As scientists chip away at the complex communication systems of animals, they move closer to understanding what creatures are saying—and maybe even how to talk back. But as we try to bridge the linguistic gap between humans and animals, some experts are raising valid concerns about whether such capabilities are appropriate—or whether we should even attempt to communicate with animals at all.

Using AI to untangle animal language

Towards the front of the pack—or should I say pod?—is Project CETI, which has used machine learning to analyze more than 8,000 sperm whale “codas”—structured click patterns recorded by the Dominica Sperm Whale Project. Researchers uncovered contextual and combinatorial structures in the whales’ clicks, naming features like “rubato” and “ornamentation” to describe how whales subtly adjust their vocalizations during conversation. These patterns helped the team create a kind of phonetic alphabet for the animals—an expressive, structured system that may not be language as we know it but reveals a level of complexity that researchers weren’t previously aware of. Project CETI is also working on ethical guidelines for the technology, a critical goal given the risks of using AI to “talk” to the animals.

Meanwhile, Google and the Wild Dolphin Project recently introduced DolphinGemma, a large language model (LLM) trained on 40 years of dolphin vocalizations. Just as ChatGPT is an LLM for human inputs—taking visual information like research papers and images and producing responses to relevant queries—DolphinGemma intakes dolphin sound data and predicts what vocalization comes next. DolphinGemma can even generate dolphin-like audio, and the researchers’ prototype two-way system, Cetacean Hearing Augmentation Telemetry (fittingly, CHAT), uses a smartphone-based interface that dolphins employ to request items like scarves or seagrass—potentially laying the groundwork for future interspecies dialogue.

“DolphinGemma is being used in the field this season to improve our real-time sound recognition in the CHAT system,” said Denise Herzing, founder and director of the Wild Dolphin Project, which spearheaded the development of DolphinGemma in collaboration with researchers at Google DeepMind, in an email to Gizmodo. “This fall we will spend time ingesting known dolphin vocalizations and let Gemma show us any repeatable patterns they find,” such as vocalizations used in courtship and mother-calf discipline.

In this way, Herzing added, the AI applications are two-fold: Researchers can use it both to explore dolphins’ natural sounds and to better understand the animals’ responses to human mimicking of dolphin sounds, which are synthetically produced by the AI CHAT system.

Expanding the animal AI toolkit

Outside the ocean, researchers are finding that human speech models can be repurposed to decode terrestrial animal signals, too. A University of Michigan-led team used Wav2Vec2—a speech recognition model trained on human voices—to identify dogs’ emotions, genders, breeds, and even individual identities based on their barks. The pre-trained human model outperformed a version trained solely on dog data, suggesting that human language model architectures could be surprisingly effective in decoding animal communication.

Of course, we need to consider the different levels of sophistication these AI models are targeting. Determining whether a dog’s bark is aggressive or playful, or whether it’s male or female—these are perhaps understandably easier for a model to determine than, say, the nuanced meaning encoded in sperm whale phonetics. Nevertheless, each study inches scientists closer to understanding how AI tools, as they currently exist, can be best applied to such an expansive field—and gives the AI a chance to train itself to become a more useful part of the researcher’s toolkit.

And even cats—often seen as aloof—appear to be more communicative than they let on. In a 2022 study out of Paris Nanterre University, cats showed clear signs of recognizing their owner’s voice, but beyond that, the felines responded more intensely when spoken to directly in “cat talk.” That suggests cats not only pay attention to what we say, but also how we say it—especially when it comes from someone they know.

Earlier this month, a pair of cuttlefish researchers found evidence that the animals have a set of four “waves,” or physical gestures, that they make to one another, as well as to human playback of cuttlefish waves. The group plans to apply an algorithm to categorize the types of waves, automatically track the creatures’ movements, and understand the contexts in which the animals express themselves more rapidly.

Private companies (such as Google) are also getting in on the act. Last week, China’s largest search engine, Baidu, filed a patent with the country’s IP administration proposing to translate animal (specifically cat) vocalizations into human language. The quick and dirty on the tech is that it would intake a trove of data from your kitty, and then use an AI model to analyze the data, determine the animal’s emotional state, and output the apparent human language message your pet was trying to convey.

A universal translator for animals?

Together, these studies represent a major shift in how scientists are approaching animal communication. Rather than starting from scratch, research teams are building tools and models designed for humans—and making advances that would have taken much longer otherwise. The end goal could (read: could) be a kind of Rosetta Stone for the animal kingdom, powered by AI.

“We’ve gotten really good at analyzing human language just in the last five years, and we’re beginning to perfect this practice of transferring models trained on one dataset and applying them to new data,” said Sara Keen, a behavioral ecologist and electrical engineer at the Earth Species Project, in a video call with Gizmodo.

The Earth Species Project plans to launch its flagship audio-language model for animal sounds, NatureLM, this year, and a demo for NatureLM-audio is already live. With input data from across the tree of life—as well as human speech, environmental sounds, and even music detection—the model aims to become a converter of human speech into animal analogues. The model “shows promising domain transfer from human speech to animal communication,” the project states, “supporting our hypothesis that shared representations in AI can help decode animal languages.”

“A big part of our work really is trying to change the way people think about our place in the world,” Keen added. “We’re making cool discoveries about animal communication, but ultimately we’re finding that other species are just as complicated and nuanced as we are. And that revelation is pretty exciting.”

The ethical dilemma

Indeed, researchers generally agree on the promise of AI-based tools for improving the collection and interpretation of animal communication data. But some feel that there’s a breakdown in communication between that scholarly familiarity and the public’s perception of how these tools can be applied.

“I think there’s currently a lot of misunderstanding in the coverage of this topic—that somehow machine learning can create this contextual knowledge out of nothing. That so long as you have thousands of hours of audio recordings, somehow some magic machine learning black box can squeeze meaning out of that,” said Christian Rutz, an expert in animal behavior and cognition and founding president of International Bio-Logging Society, in a video call with Gizmodo. “That’s not going to happen.”

“Meaning comes through the contextual annotation and this is where I think it’s really important for this field as a whole, in this period of excitement and enthusiasm, to not forget that this annotation comes from basic behavioral ecology and natural history expertise,” Rutz added. In other words, let’s not put the horse before the cart, especially since the cart—in this case—is what’s powering the horse.

But with great power…you know the cliché. Essentially, how can humans develop and apply these technologies in a way that is both scientifically illuminating and minimizes harm or disruption to its animal subjects? Experts have put forward ethical standards and guardrails for using the technologies that prioritize the welfare of creatures as we get closer to—well, wherever the technology is going.

As AI advances, conversations about animal rights will have to evolve. In the future, animals could become more active participants in those conversations—a notion that legal experts are exploring as a thought exercise, but one that could someday become reality.

“What we desperately need—apart from advancing the machine learning side—is to forge these meaningful collaborations between the machine learning experts and the animal behavior researchers,” Rutz said, “because it’s only when you put the two of us together that you stand a chance.”

There’s no shortage of communication data to feed into data-hungry AI models, from pitch-perfect prairie dog squeaks to snails’ slimy trails (yes, really). But exactly how we make use of the information we glean from these new approaches requires thorough consideration of the ethics involved in “speaking” with animals.

A recent paper on the ethical concerns of using AI to communicate with whales outlined six major problem areas. These include privacy rights, cultural and emotional harm to whales, anthropomorphism, technological solutionism (an overreliance on technology to fix problems), gender bias, and limited effectiveness for actual whale conservation. That last issue is especially urgent, given how many whale populations are already under serious threat.

It increasingly appears that we’re on the brink of learning much more about the ways animals interact with one another—indeed, pulling back the curtain on their communication could also yield insights into how they learn, socialize, and act within their environments. But there are still significant challenges to overcome, such as asking ourselves how we use the powerful technologies currently in development.

Becoming a centaur (Aeon)

Rounding up wild horses on the edge of the Gobi desert in Mongolia, 1964. Photo by Philip Jones Griffiths/Magnum
The horse is a prey animal, the human a predator. Our shared trust and athleticism is a neurobiological miracle

Janet Jones – 14 January 2022

Horse-and-human teams perform complex manoeuvres in competitions of all sorts. Together, we can gallop up to obstacles standing 8 feet (2.4 metres) high, leave the ground, and fly blind – neither party able to see over the top until after the leap has been initiated. Adopting a flatter trajectory with greater speed, horse and human sail over broad jumps up to 27 feet (more than 8 metres) long. We run as one at speeds of 44 miles per hour (nearly 70 km/h), the fastest velocity any land mammal carrying a rider can achieve. In freestyle dressage events, we dance in place to the rhythm of music, trot sideways across the centre of an arena with huge leg-crossing steps, and canter in pirouettes with the horse’s front feet circling her hindquarters. Galloping again, the best horse-and-human teams can slide 65 feet (nearly 20 metres) to a halt while resting all their combined weight on the horse’s hind legs. Endurance races over extremely rugged terrain test horses and riders in journeys that traverse up to 500 miles (805 km) of high-risk adventure.

Charlotte Dujardin on Valegro, a world-record dressage freestyle at London Olympia, 2014: an example of high-precision brain-to-brain communication between horse and rider. Every step the horse takes is determined in conjunction with many invisible cues from his human rider, using a feedback loop between predator brain and prey brain. Note the horse’s beautiful physical condition and complete willingness to perform these extremely difficult manoeuvres.

No one disputes the athleticism fuelling these triumphs, but few people comprehend the mutual cross-species interaction that is required to accomplish them. The average horse weighs 1,200 pounds (more than 540 kg), makes instantaneous movements, and can become hysterical in a heartbeat. Even the strongest human is unable to force a horse to do anything she doesn’t want to do. Nor do good riders allow the use of force in training our magnificent animals. Instead, we hold ourselves to the higher standard of motivating horses to cooperate freely with us in achieving the goals of elite sports as well as mundane chores. Under these conditions, the horse trained with kindness, expertise and encouragement is a willing, equal participant in the action.

That action is rooted in embodied perception and the brain. In mounted teams, horses, with prey brains, and humans, with predator brains, share largely invisible signals via mutual body language. These signals are received and transmitted through peripheral nerves leading to each party’s spinal cord. Upon arrival in each brain, they are interpreted, and a learned response is generated. It, too, is transmitted through the spinal cord and nerves. This collaborative neural action forms a feedback loop, allowing communication from brain to brain in real time. Such conversations allow horse and human to achieve their immediate goals in athletic performance and everyday life. In a very real sense, each species’ mind is extended beyond its own skin into the mind of another, with physical interaction becoming a kind of neural dance.

Horses in nature display certain behaviours that tempt observers to wonder whether competitive manoeuvres truly require mutual communication with human riders. For example, the feral horse occasionally hops over a stream to reach good food or scrambles up a slope of granite to escape predators. These manoeuvres might be thought the precursors to jumping or rugged trail riding. If so, we might imagine that the performance horse’s extreme athletic feats are innate, with the rider merely a passenger steering from above. If that were the case, little requirement would exist for real-time communication between horse and human brains.

In fact, though, the feral hop is nothing like the trained leap over a competition jump, usually commenced from short distances at high speed. Today’s Grand Prix jump course comprises about 15 obstacles set at sharp angles to each other, each more than 5 feet high and more than 6 feet wide (1.5 x 1.8 metres). The horse-and-human team must complete this course in 80 or 90 seconds, a time allowance that makes for acute turns, diagonal flight paths and high-speed exits. Comparing the wilderness hop with the show jump is like associating a flintstone with a nuclear bomb. Horses and riders undergo many years of daily training to achieve this level of performance, and their brains share neural impulses throughout each experience.

These examples originate in elite levels of horse sport, but the same sort of interaction occurs in pastures, arenas and on simple trails all over the world. Any horse-and-human team can develop deep bonds of mutual trust, and learn to communicate using body language, knowledge and empathy.

Like it or not, we are the horse’s evolutionary enemy, yet they behave toward us as if inclined to become a friend

The critical component of the horse in nature, and her ability to learn how to interact so precisely with a human rider, is not her physical athleticism but her brain. The first precise magnetic resonance image of a horse’s brain appeared only in 2019, allowing veterinary neurologists far greater insight into the anatomy underlying equine mental function. As this new information is disseminated to horse trainers and riders for practical application, we see the beginnings of a revolution in brain-based horsemanship. Not only will this revolution drive competition to higher summits of success, and animal welfare to more humane levels of understanding, it will also motivate scientists to research the unique compatibility between prey and predator brains. Nowhere else in nature do we see such intense and intimate collaboration between two such disparate minds.

Three natural features of the equine brain are especially important when it comes to mind-melding with humans. First, the horse’s brain provides astounding touch detection. Receptor cells in the horse’s skin and muscles transduce – or convert – external pressure, temperature and body position to neural impulses that the horse’s brain can understand. They accomplish this with exquisite sensitivity: the average horse can detect less pressure against her skin than even a human fingertip can.

Second, horses in nature use body language as a primary medium of daily communication with each other. An alpha mare has only to flick an ear toward a subordinate to get him to move away from her food. A younger subordinate, untutored in the ear flick, receives stronger body language – two flattened ears and a bite that draws blood. The notion of animals in nature as kind, gentle creatures who never hurt each other is a myth.

Third, by nature, the equine brain is a learning machine. Untrammelled by the social and cognitive baggage that human brains carry, horses learn in a rapid, pure form that allows them to be taught the meanings of various human cues that shape equine behaviour in the moment. Taken together, the horse’s exceptional touch sensitivity, natural reliance on body language, and purity of learning form the tripod of support for brain-to-brain communication that is so critical in extreme performance.

One of the reasons for budding scientific fascination with neural horse-and-human communication is the horse’s status as a prey animal. Their brains and bodies evolved to survive completely different pressures than our human physiologies. For example, horse eyes are set on either side of their head for a panoramic view of the world, and their horizontal pupils allow clear sight along the horizon but fuzzy vision above and below. Their eyes rotate to maintain clarity along the horizon when their heads lie sideways to reach grass in odd locations. Equine brains are also hardwired to stream commands directly from the perception of environmental danger to the motor cortex where instant evasion is carried out. All of these features evolved to allow the horse to survive predators.

Conversely, human brains evolved in part for the purpose of predation – hunting, chasing, planning… yes, even killing – with front-facing eyes, superb depth perception, and a prefrontal cortex for strategy and reason. Like it or not, we are the horse’s evolutionary enemy, yet they behave toward us as if inclined to become a friend.

The fact that horses and humans can communicate neurally without the external mediation of language or equipment is critical to our ability to initiate the cellular dance between brains. Saddles and bridles are used for comfort and safety, but bareback and bridleless competitions prove they aren’t necessary for highly trained brain-to-brain communication. Scientific efforts to communicate with predators such as dogs and apes have often been hobbled by the use of artificial media including human speech, sign language or symbolic lexigram. By contrast, horses allow us to apply a medium of communication that is completely natural to their lives in the wild and in captivity.

The horse’s prey brain is designed to notice and evade predators. How ironic, and how riveting, then, that this prey brain is the only one today that shares neural communication with a predator brain. It offers humanity a rare view into a prey animal’s world, almost as if we were wolves riding elk or coyotes mind-melding with cottontail bunnies.

Highly trained horses and riders send and receive neural signals using subtle body language. For example, a rider can apply invisible pressure with her left inner calf muscle to move the horse laterally to the right. That pressure is felt on the horse’s side, in his skin and muscle, via proprioceptive receptor cells that detect body position and movement. Then the signal is transduced from mechanical pressure to electrochemical impulse, and conducted up peripheral nerves to the horse’s spinal cord. Finally, it reaches the somatosensory cortex, the region of the brain responsible for interpreting sensory information.

Riders can sometimes guess that an invisible object exists by detecting subtle equine reactions

This interpretation is dependent on the horse’s knowledge that a particular body signal – for example, inward pressure from a rider’s left calf – is associated with a specific equine behaviour. Horse trainers spend years teaching their mounts these associations. In the present example, the horse has learned that this particular amount of pressure, at this speed and location, under these circumstances, means ‘move sideways to the right’. If the horse is properly trained, his motor cortex causes exactly that movement to occur.

By means of our human motion and position sensors, the rider’s brain now senses that the horse has changed his path rightward. Depending on the manoeuvre our rider plans to complete, she will then execute invisible cues to extend or collect the horse’s stride as he approaches a jump that is now centred in his vision, plant his right hind leg and spin in a tight fast circle, push hard off his hindquarters to chase a cow, or any number of other movements. These cues are combined to form that mutual neural dance, occurring in real time, and dependent on natural body language alone.

The example of a horse moving a few steps rightward off the rider’s left leg is extremely simplistic. When you imagine a horse and rider clearing a puissance wall of 7.5 feet (2.4 metres), think of the countless receptor cells transmitting bodily cues between both brains during approach, flight and exit. That is mutual brain-to-brain communication. Horse and human converse via body language to such an extreme degree that they are able to accomplish amazing acts of understanding and athleticism. Each of their minds has extended into the other’s, sending and receiving signals as if one united brain were controlling both bodies.

Franke Sloothaak on Optiebeurs Golo, a world-record puissance jump at Chaudfontaine in Belgium, 1991. This horse-and-human team displays the gentle encouragement that brain-to-brain communication requires. The horse is in perfect condition and health. The rider offers soft, light hands, and rides in perfect balance with the horse. He carries no whip, never uses his spurs, and employs the gentlest type of bit – whose full acceptance is evidenced by the horse’s foamy mouth and flexible neck. The horse is calm but attentive before and after the leap, showing complete willingness to approach the wall without a whiff of coercion. The first thing the rider does upon landing is pat his equine teammate. He strokes or pats the horse another eight times in the next 30 seconds, a splendid example of true horsemanship.

Analysis of brain-to-brain communication between horses and humans elicits several new ideas worthy of scientific notice. Because our minds interact so well using neural networks, horses and humans might learn to borrow neural signals from the party whose brain offers the highest function. For example, horses have a 340-degree range of view when holding their heads still, compared with a paltry 90-degree range in humans. Therefore, horses can see many objects that are invisible to their riders. Yet riders can sometimes guess that an invisible object exists by detecting subtle equine reactions.

Specifically, neural signals from the horse’s eyes carry the shape of an object to his brain. Those signals are transferred to the rider’s brain by a well-established route: equine receptor cells in the retina lead to equine detector cells in the visual cortex, which elicits an equine motor reaction that is then sensed by the rider’s human body. From there, the horse’s neural signals are transmitted up the rider’s spinal cord to the rider’s brain, and a perceptual communication loop is born. The rider’s brain can now respond neurally to something it is incapable of seeing, by borrowing the horse’s superior range of vision.

These brain-to-brain transfers are mutual, so the learning equine brain should also be able to borrow the rider’s vision, with its superior depth perception and focal acuity. This kind of neural interaction results in a horse-and-human team that can sense far more together than either party can detect alone. In effect, they share effort by assigning labour to the party whose skills are superior at a given task.

There is another type of skillset that requires a particularly nuanced cellular dance: sharing attention and focus. Equine vigilance allowed horses to survive 56 million years of evolution – they had to notice slight movements in tall grasses or risk becoming some predator’s dinner. Consequently, today it’s difficult to slip even a tiny change past a horse, especially a young or inexperienced animal who has not yet been taught to ignore certain sights, sounds and smells.

By contrast, humans are much better at concentration than vigilance. The predator brain does not need to notice and react instantly to every stimulus in the environment. In fact, it would be hampered by prey vigilance. While reading this essay, your brain sorts away the sound of traffic past your window, the touch of clothing against your skin, the sight of the masthead that says ‘Aeon’ at the top of this page. Ignoring these distractions allows you to focus on the content of this essay.

Horses and humans frequently share their respective attentional capacities during a performance. A puissance horse galloping toward an enormous wall cannot waste vigilance by noticing the faces of each person in the audience. Likewise, the rider cannot afford to miss a loose dog that runs into the arena outside her narrow range of vision and focus. Each party helps the other through their primary strengths.

Such sharing becomes automatic with practice. With innumerable neural contacts over time, the human brain learns to heed signals sent by the equine brain that say, in effect: ‘Hey, what’s that over there?’ Likewise, the equine brain learns to sense human neural signals that counter: ‘Let’s focus on this gigantic wall right here.’ Each party sends these messages by body language and receives them by body awareness through two spinal cords, then interprets them inside two brains, millisecond by millisecond.

The rider’s physical cues are transmitted by neural activation from the horse’s surface receptors to the horse’s brain

Finally, it is conceivable that horse and rider can learn to share features of executive function – the human brain’s ability to set goals, plan steps to achieve them, assess alternatives, make decisions and evaluate outcomes. Executive function occurs in the prefrontal cortex, an area that does not exist in the equine brain. Horses are excellent at learning, remembering and communicating – but they do not assess, decide, evaluate or judge as humans do.

Shying is a prominent equine behaviour that might be mediated by human executive function in well-trained mounts. When a horse of average size shies away from an unexpected stimulus, riders are sitting on top of 1,200 pounds of muscle that suddenly leaps sideways off all four feet and lands five yards away. It’s a frightening experience, and often results in falls that lead to injury or even death. The horse’s brain causes this reaction automatically by direct connection between his sensory and motor cortices.

Though this possibility must still be studied by rigorous science, brain-to-brain communication suggests that horses might learn to borrow small glimmers of executive function through neural interaction with the human’s prefrontal cortex. Suppose that a horse shies from an umbrella that suddenly opens. By breathing steadily, relaxing her muscles, and flexing her body in rhythm with the horse’s gait, the rider calms the animal using body language. Her physical cues are transmitted by neural activation from his surface receptors to his brain. He responds with body language in which his muscles relax, his head lowers, and his frightened eyes return to their normal size. The rider feels these changes with her body, which transmits the horse’s neural signals to the rider’s brain.

From this point, it’s only a very short step – but an important one – to the transmission and reception of neural signals between the rider’s prefrontal cortex (which evaluates the unexpected umbrella) and the horse’s brain (which instigates the leap away from that umbrella). In practice, to reduce shying, horse trainers teach their young charges to slow their reactions and seek human guidance.

Brain-to-brain communication between horses and riders is an intricate neural dance. These two species, one prey and one predator, are living temporarily in each other’s brains, sharing neural information back and forth in real time without linguistic or mechanical mediation. It is a partnership like no other. Together, a horse-and-human team experiences a richer perceptual and attentional understanding of the world than either member can achieve alone. And, ironically, this extended interspecies mind operates well not because the two brains are similar to each other, but because they are so different.

Janet Jones applies brain research to training horses and riders. She has a PhD from the University of California, Los Angeles, and for 23 years taught the neuroscience of perception, language, memory, and thought. She trained horses at a large stable early in her career, and later ran a successful horse-training business of her own. Her most recent book, Horse Brain, Human Brain (2020), is currently being translated into seven languages.

Edited by Pam Weintraub