If sperm whales could describe how swimming in polluted waters and living with the din of boats and bright lights interfere with communications with their family and friends and cause pain and suffering, would the court admit that testimony? Maybe someday.
Fast-moving developments in artificial intelligence could one day crack the codes to translating animals’ whistles, chirps, clicks and barks—allowing humans and animals to communicate with each other at new levels and spawning a host of legal and ethical questions.
“We need to have these conversations right now because, importantly, the law always lags technology,” says Jamie McLaughlin, a vice chair of the ABA International Animal Law Committee and an associate attorney at the Nonhuman Rights Program.
For instance, the DolphinGemma project is training a Google-developed large language model using dolphin audio to decode the marine mammal’s communications—and possibly talk back in their language. And, the Earth Species Project’s model processes sounds from various animals to potentially communicate across species.
But with that power would come responsibility.
“The use of AI and other technologies to enable or augment the ‘translation’ of nonhuman animal communications—to say nothing of enabling true bilateral interspecies communication—poses as many risks as it does opportunities,” says César Rodríguez-Garavito, the director of the More Than Human Life Program, or MOTH, and a professor at the New York University School of Law.
These developments beg a host of legal questions including:
Training animals, especially relatively big-brained, charismatic animals like dolphins, whales and chimpanzees, how to communicate with humans has been going on for decades.
For instance, Koko, a chimpanzee, was taught over 1,000 signs and seemed to understand about 2,000 spoken English words language, according to the Gorilla Foundation.
“We wanted them to speak in our language. but that's not natural to the animals,” says Rajesh K. Reddy, assistant professor at Lewis & Clark Law School and director of its animal law program.
“We are understanding them through our own words partly because it's quite difficult to do anything else,” says Kathy Hessler, assistant dean for animal law at the George Washington University Law School, and director of the Animal Legal Education Initiative, adding that this communication is based on “human exceptionalism.”
But AI might change that. Recent findings show many animals have sophisticated communications, and sperm whales even have a phonetic alphabet.
“What AI allows us to do is pick up the noises that we can't hear, process that to an extent that we cannot even fathom and determine what these nonhuman animals are saying to one another,” Reddy says. “AI could really unlock our ability to say what these particular species and individual animals are really experiencing.”
While AI's capabilities to translate animal languages are still in their nascent phase, some AI researchers are already collaborating with legal experts.
Project CETI—Cetacean Translation Initiation—is a listening project bringing together scientists in biology, linguistics and robotics using machine learning to translate sperm whales’ clicks and songs. Well aware of potential legal and ethical issues, CETI collaborates with the MOTH Program, which researches the legal impact of AI-assisted studies of animal communication.
Over decades of work, Project CETI has found sperm whales have conscious thoughts and can plan for the future while experiencing a range of emotions, says MOTH Program’s Rodríguez-Garavito, author of “What If We Understood What Animals Are Saying? The Legal Impact Of AI-assisted Studies of Animal Communication,” published in Ecology Law Quarterly in May.
For instance, chronic vessel noise can compromise whales’ ability to detect, recognize or understand each other's sounds regarding feeding, navigation and mating—communication with potentially grave consequences, says Ashley Nemeth, supervising attorney at the MOTH Program.
But could translated whale communications be admitted as evidence?
“It’s hard not to give a stereotypically lawyerly answer that ‘it depends,’” says MOTH’s Nemeth.
“We need to have these conversations right now because, importantly, the law always lags technology.”
Reddy agrees. “If we are able to understand how an animal who has been harmed, that could be really meaningful in terms of how a judge or jury sees and appreciates what justice for that animal might mean,” he says.
While potentially years away, that type of information would be assessed by the court like any other scientific evidence, sources say.
“We need to answer this fundamental scientific question: ‘Is the science able to meet our traditional Daubert standard?’” asks Hessler. “How do we know when we have enough data that has been peer reviewed and is substantial enough to count as expert understanding?”
And how the AI translator is trained and by whom would be important to understand its bias, McLaughlin says.
“International shipping companies, for example, might want to keep their boat speeds and not want the expense of refitting with quieter engines,” she says. “They might put forth other data that says, ‘That's not what these whales are saying.’
“It's the person with the better resources that wins,” she adds.
In recent years, court cases like Justice v. Vercher have held that animals lack legal standing.
But improved communications via AI could change that, some say.
"What the courts have cared about before was that we don't know what they're thinking," says Robyn Pekala, an animal rights attorney and a vice chair of the ABA TIPS Animal Law Committee.
But Yolanda Eisenstein, president of Paris-based Union Internationale des Avocat's Animal Law Commision, says it's too soon to have this conversation.
"It's concerning to me to have this idea of AI and communications somehow being able to change that or solve that," says the winner of the ABA's 2018 Excellence in the Advancement of Animal Law Award. "These are not just communication issues."
Being on the stand requires answering specific questions, demanding reliable two-way communications, sources say, but how close we are to a back-and-forth conversation “depends on how one defines two-way communications,” David Gruber, Project CETI founder, and Gašper Beguš, linguistic lead at Project CETI, wrote to the ABA Journal.
“More simple two-way communications could be possible in a matter of years; more complex two-way communications is a more open question," they wrote.
Even if two-way communication were possible, these question-and-answer sessions could go down a rabbit hole of ethics and legal issues, sources say.
"It could backfire. What if we find out that these animals are much less intelligent than we thought?" Pekala asks.
Also, some species might not have languages for human concepts, Hessler says.
“Constructs that we take for granted, like time, may operate differently in different species than we understand it,” Hessler says. “A bee sees color differently than a human does. So how do we how do we even talk about color?”
And there is the issue of how to determine ethical behavior, sources say. “We don't even know if their ethics align with ours,” McLaughlin says. “Maybe one animal wants retribution and another animal does not.
And what if humans offer bad information to the animals?
“We're going to have a whole bunch of people trying to chat with whales in ways that we know will be imprecise,” Hessler says. “If we end up miscommunicating, we might be saying to follow me to a dangerous place, and that could be really harmful, right?”
In addition, consent to be a witness “would itself be normatively and practically challenging,” Nemeth adds.
In March, People for the Ethical Treatment of Animal sued the National Institute of Mental Health and the National Institutes of Health, arguing that humans’ First Amendment right to free speech includes the right to receive communications from willing speakers, including macaque monkeys being tested in the government’s labs.
There’s another angle: If AI can prove that animals can’t communicate vital information to each other when a major predator known as a human is around, “there might be ways to call it the right to privacy,” Lewis & Clark Law’s Reddy says.
How this would be resolved remains an open question. "We can't even get legislation for humans on AI," Eisenstein says.
And data privacy rights for people are few. "We're recorded all the time,” McLaughlin says. “We couldn't expect anything better for animals, so it's probably going to be worse.”
Legal guardrails will be necessary, sources say. This fall, the MOTH Program will publish a set of 12 legal and ethical principles for scientists and technologists to ensure animal communication technologies are respectfully deployed, says Rodríguez-Garavito.
But some legal remedies could lean on current laws, Nemeth says. “We don’t necessarily need to create new legal paradigms to act on groundbreaking science," she adds. "We can begin by using the laws we already have to better protect marine mammals."
Under the Endangered Species Act and Marine Mammal Protection Act, Nemeth argues, translated cetacean communication could encourage federal agencies to regulate vessel noise and potentially help litigants demonstrate irreparable harm since that constant overstimulation could be equated to similar human torture methods.
“The right to be free from torture and cruel treatment is perhaps one of the most enduring and widespread international legal principles that exists,” says Nemeth, and it mandates redress.
Other logistical questions remain, sources say, including should an animal physically appear in court and how they would they find a lawyer.
“Who is going to be representing the animal in court? Is it going to be a guardian ad litem situation? Is it going to be a next friend status?” McLaughlin says. “How do we know that they have the best interest of the animal?”
As AI’s capabilities continue to further evolve, “we need to figure out how to address the downsides, the ethical issues, the welfare implications for the animals,” Hessler says.
“We have to step back and ask, why are we engaging in this work, and to whose benefit? It's always going to be partially, at least partially human benefit, or we wouldn't be doing it," she adds. “But this conversation is here, and we have the social reality of it, the technical reality of it, and the ethics and legal reality too.”
