Look Who’s Talking
It’s the animals, and they’re very upset
Leo Kim/THE BAFFLER MAGAZINE
November 18, 2024
DURING THE HEIGHT OF THE PANDEMIC, when life was reduced to our screens, a poodle named Bunny captivated the world by speaking. More accurately, she captivated us by pressing buttons on a DIY soundboard that played a range of preprogrammed words. At first, Bunny had simple requests: a walk here, a treat there. Over time, however, Bunny’s ideas appeared to become more complex. In one video, she faced a mirror and button-mashed the question, “What, Bunny?” Many found themselves asking the same thing. Some saw her as a circus act for the TikTok age, while others saw a creature attempting to genuinely express itself.
The Seal of Solomon was once said to have granted the Israelite king the ability to speak to animals. These days, we have less faith in scriptural abracadabra; yet in their stead, a new magic has emerged: technology. Since Bunny first said hello to the world, this magic has only grown in scale and complexity. Just a few years later and that soundboard already appears like a primitive relic of a distant past, gone the way of Solomon. Now it’s no longer low-tech buttons but sleek “artificial intelligence” — that newest and most powerful of our sacred technologies — that has captured our imagination and hopes of interspecies communication.
Having successfully translated English to French and French to Russian and Russian to Japanese, researchers are now interested in using neural networks to translate Whale to English, Chimpanzee to French. The proposal may sound like the stuff of fairy tales, but the idea has gained traction amid a hype-driven mediascape in which just about anything slapped with the AI label is considered of potentially epochal importance. Buzzy headlines have started to speculate how “Artificial Intelligence Could Finally Let Us Talk with Animals,” questioned “How First Contact With Whale Civilization Could Unfold,” and giddily anticipated the “AI Pet Translator Apps” to come. In the future, Bunny will be legion.
Lest one accuse this project of superfluity, its proponents have bestowed a certain moral urgency on the effort. Using technology to help us hear these creatures out and witness their plight, the thinking goes, will shake us from the murderous apathy that has plagued our relationship with the natural world. Though past attempts at appealing to the better angels of our nature haven’t exactly stopped the slaughter, these techno-solutionists assure us that AI will allow us to resolve this problem of miscommunication once and for all by letting us hear straight from the literal horse’s mouth. We won’t be able to brush off the desperate pleas of these creatures any longer because they’ll be begging for their lives in fluent English.
The notion that speaking with animals might deliver us to some deeper empathic attunement with the natural world is relatively new. For most of Western history, we assumed that animals simply didn’t have anything interesting to say. Two millennia ago, Aristotle insisted that logos — the capacity for reflection, reason, and language — separated man from the rest of the animal world. During the first throes of the Enlightenment, Descartes, who had constructed his entire worldview off of the singular certainty of the thinking mind, described nonhuman animals as thoughtless meat machines reacting impulsively to stimuli. Even at the end of the nineteenth century, mainstream anthropologists thought that humans might have arisen from a “revolution” in cognition that ruptured us from the mute animal world.
All that now separates us from enlightenment is a few more rounds of funding.
Only in the past century did we begin to earnestly explore the possibility that animals might suffer and hope and feel. The 1960s and 1970s in particular saw a surge of interest in animal cognition and communication. It was during this time that researchers taught chimps like Washoe sign language and worked to decode the complex signals of species like songbirds and dolphins. Even considering the more gruesome experiments carried out in this consciousness-expanding spirit (honorable mention goes to the researchers that killed an elephant by injecting it with 297 mg of LSD and the Navy project to enlist dolphins in the fight against godless communism), this period is generally remembered for countering our dismissive view of the animal kingdom. We started to see these creatures as more than mindless automata. When we learned that whales sing, for instance, we played their songs to galvanize public support against whaling.
Inspired by this history, contemporary techno-solutionists have figured that we might bring this arc of empathy to its zenith by tearing down the final barrier they see as separating us: language. There are now research initiatives looking to decode everything from elephant calls to the dancing movements of bees and melodies of birdsong, all inspired by this notion that piecing together the fragments of Babel and restoring communication between earth’s creatures will finally usher in an era of harmony. These tools will help “a society that is disconnected from nature to reconnect with it,” as one leading researcher told Wired. Or as the highly publicized Cetacean Translation Initiative, or Project CETI for short, put it, if the discovery that whales could sing helped ban whaling, then just “imagine what would happen if we could understand” them.
The sentiment has opened wallets. Project CETI has raised upwards of $33 million for their plan, which they lay out in three easy steps:
1. Achieve advancements and breakthroughs in interspecies communication.
2. Transform human understanding & connectivity.
3. Share our learnings with the world.
As the background of the team — composed largely of machine-learning researchers from both academia and for-profit companies like Google — makes clear, these “advances and breakthroughs” are expected to come from the domain of machine learning: an application of AI focused on developing systems that autonomously refine their performance over time. In its simplest terms, these technologies would map the structure of the whale-speak, plotting different linguistic units within a multidimensional matrix of relationships; then, by overlaying the complex “shape” of whale language to that of our own, researchers could translate between the two without the need of a mediating Rosetta Stone since overlapping points on these maps would indicate a corresponding definition. From there, the transformation of “human understanding and connectivity” would follow, obviously.
Similar enterprises have sprung up across the board, looking to get in on the action. Apps like MeowTalk and such organizations as Zoolingua have set their sights on our furry friends — using AI to mediate our relationships with our domesticated cats and dogs, respectively. Meanwhile more grandiose (and better funded) organizations like the Earth Species Project have projects underway to help decode the communication of crows and beluga whales, among others. “Like the telescope, these new tools will forever change our perspective,” they proclaim. Though one might reasonably question whether these “new tools” will actually resemble the telescope as much as the parade of pernicious platforms and technologies that Earth Species’ cofounders — which include a founding member of the platform formerly known as Twitter, the creator of the infinite scroll, and “an early member” of the Facebook team — have had a hand in developing in the past.
Regardless, these plans continue to gleam with the confidence of a syllogism. Having conceptually intertwined technological innovation with empathic transformation, all that now separates us from enlightenment is a few more rounds of funding. Yet the foundations underlying this dream of tech enabled eco-communion are dubious at best. In fact, subjected to even the mildest of scrutiny, it becomes clear that this salvific faith in technology betrays less a commitment to transforming ourselves for the better than it does a lazy desire to double down on the same traits that landed us in this mess: a persistent belief in the power of individual reason, an overreliance on techno-fixes, and a constant need to shape the world in our own image.
Given that this whole scheme relies on the transformative power of language, it’s surprising that the perils that surround this power are so often ignored — a consequence, perhaps, of the fact that these projects rarely employ more than a handful of actual linguists. Language is not the blindly unifying tool that these techno-delusionists characterize it as; it not only builds bridges but burns them down. My grandfather grew up learning Japanese because his native tongue, Korean, had been banned by the colonial government. It’s a familiar story for anyone who has suffered under imperialism, whether you’re Korean, Cherokee, Algerian, Irish, (the list goes on). Language contains and constructs worlds, serves as the vehicle through which culture disseminates and politics arise. To control language is to control the medium that conditions thought. No wonder that the forced imposition of language is a favorite tactic of empire.
These initiatives are little more than red herrings that, like most techno-fixes, puts the onus of change on individual self-betterment instead of systemic upheaval.
Seen in this light, the desire to translate nonhuman thought into human language is anything but innocuous. The philosopher of language Ludwig Wittgenstein once quipped that “if a lion could talk, we could not understand him.” This aphorism is a reminder that the umwelten of these animals — the phenomenal shape of their experience, the mental states that orient them toward the world — are so radically different from our own, the language we have crafted for our all too human lives cannot capture the full breadth of their existence. It’s one reason why, as a recent New Yorker article observed, many animal communications experts don’t expect “straightforward translations” to be forthcoming, since “animals live in perceptual worlds that are just too different.” Eco-theorists have built on this, arguing that an honest relationship with the nonhuman requires us to first acknowledge the fundamental differences that shape us. Only by cultivating an empathy that cuts across these divides might we move beyond the narcissism that has long polluted our relationship to the broader world.
Yet rather than help us appreciate this radical diversity and overcome our self-centeredness, AI translation aims to smooth everything out into easily legible conformity. This wouldn’t deliver us to real understanding as much as its false pretense — turning bats, lions, and whales into cheap caricatures robbed of complexity. Conveniently, this would also allow us to avoid the difficult task of teaching ourselves to see and listen to these animals in a way that respects all that’s unique about them. Why stretch our networks of care to these alien forms of existence when we can just reinforce the self-serving notion that life only matters to the extent that it is “like us?” Having confronted something wonderful and new, we’ve opted not to explore these vast unknowns — by crafting new concepts or modes of experience capable of capturing the unfamiliar — but have told them to speak English.
Despite these concerns, there will be those who continue to believe in the possibility of meaningful translation. For these techno-optimists’ sake, let’s assume that everything miraculously works out. Let’s assume that we’ll successfully extend our already bloated surveillance apparatuses into the oceans and forests to scrape enough data to feed these models — and that the shapes of these animal and human signaling systems will line up so perfectly that these creatures become our new English-speaking, all-American pen pals. Even in this best case scenario it’s still unclear what we would say to each other, much less what this would actually achieve.
After all, it’s not like we’re utterly incapable of understanding animals right now. The idea that we might use this line of communication to ask whales if shipping vessels have grown too loud or if they can feel the effects of climate change, as some have proposed, betrays an implausible naivete. Of course our boats are too loud; of course whales can feel the effects of climate change. Their suffering is no great mystery — just look at how they die, starving with bellies full of plastic. AI couldn’t do more to convey what is already so obvious to anyone willing to look. We’re not destroying the planet because we lack the right “facts of the matter” but because our devastating systems are fundamentally irrational, premised on paradoxical axioms like perpetual growth.
Not to say that versions of these tools don’t have useful local implementations. Author and researcher Karen Bakker, for instance, has observed how shipping lanes can respond to whale traffic by employing digital bioacoustics and predictive AI to identify the paths of these creatures and reroute ships accordingly. Yet absent is the sense that global winds of consumption blowing these ships across oceans (and straight into twenty thousand whales annually) will change from speaking to these animals. Though we might prevent a few hit-and-runs, a multimillion dollar equivalent of a school crossing guard is a far cry from the transformative panacea that this technology is being hyped up to be.
If these technologies do hold the appeal of total transformation, the nature of this transformation lies elsewhere. We sense it when, for example, considering the question of what we might say to these animals, the late biologist Roger Payne remarked that “sorry would be a good word.” Tellingly, it’s a sentiment frequently echoed in conversations about this research. In the end, we seem to desire purification through the sacrament of confession more than hearing what these creatures would say in order to actually act on it. We’ve convinced ourselves that saving these animals first requires us to save ourselves by assuaging our own guilt. In short, as we so often do, we’ve made it all about us.
These initiatives are little more than red herrings that, like most techno-fixes, puts the onus of change on individual self-betterment instead of systemic upheaval. Yet no amount of feel-good personal enlightenment will alter anything without first dismantling the economic and legal infrastructures desecrating the planet. The romantic role of whale song in banning whaling shouldn’t overshadow the mundane fact that overhunting and the invention of resource alternatives like kerosene and vegetable oils had already made whaling a far less profitable prospect by that point. Hearts may bleed, but until wallets do, few things tend to change.
Like the investor pitches of so many failed startups, the promise of AI animal communication sells us transcendence while delivering something that already exists wrapped up in silicon packaging — this time with the added irony that we’re being asked to funnel millions of dollars into a water-guzzling, emissions-pumping technology to save nature by teaching us basic empathy. If we continue to wait for the machine to equip us with the capacity to care, it’ll be too late. By the time we’ve figured out how to apologize to the whales, their songs will have long fallen silent.
