Massive benefits of AI come with environmental and human costs. Can AI itself be part of the solution?
The recent explosion of generative artificial intelligence tools like ChatGPT and Dall-E enabled anyone with internet access to harness AI’s power for enhanced productivity, creativity, and problem-solving. With their ever-improving capabilities and expanding user base, these tools proved useful across disciplines, from the creative to the scientific.
But beneath the technological wonders of human-like conversation and creative expression lies a dirty secret—an alarming environmental and human cost. AI has an immense carbon footprint. Systems like ChatGPT take months to train in high-powered data centers, which demand huge amounts of electricity, much of which is still generated with fossil fuels, as well as water for cooling. “One of the reasons why Open AI needs investments [to the tune of] $10 billion from Microsoft is because they need to pay for all of that computation,” says Kentaro Toyama, a computer scientist at the University of Michigan. There’s also an ecological toll from mining rare minerals required for hardware and infrastructure. This environmental exploitation pollutes land, triggers natural disasters and causes large-scale human displacement. Finally, for data labeling needed to train and correct AI algorithms, the Big Data industry employs cheap and exploitative labor, often from the Global South.
Generative AI tools are based on large language models (LLMs), with most well-known being various versions of GPT. LLMs can perform natural language processing, including translating, summarizing and answering questions. They use artificial neural networks, called deep learning or machine learning. Inspired by the human brain, neural networks are made of millions of artificial neurons. “The basic principles of neural networks were known even in the 1950s and 1960s,” Toyama says, “but it’s only now, with the tremendous amount of compute power that we have, as well as huge amounts of data, that it’s become possible to train generative AI models.”
Though there aren’t any official figures about the power consumption or emissions from data centers, experts estimate that they use one percent of global electricity—more than entire countries.
In recent months, much attention has gone to the transformative benefits of these technologies. But it’s important to consider that these remarkable advances may come at a price.
AI’s carbon footprint
In their latest annual report, 2023 Landscape: Confronting Tech Power, the AI Now Institute, an independent policy research entity focusing on the concentration of power in the tech industry, says: “The constant push for scale in artificial intelligence has led Big Tech firms to develop hugely energy-intensive computational models that optimize for ‘accuracy’—through increasingly large datasets and computationally intensive model training—over more efficient and sustainable alternatives.”
Though there aren’t any official figures about the power consumption or emissions from data centers, experts estimate that they use one percent of global electricity—more than entire countries. In 2019, Emma Strubell, then a graduate researcher at the University of Massachusetts Amherst, estimated that training a single LLM resulted in over 280,000 kg in CO2 emissions—an equivalent of driving almost 1.2 million km in a gas-powered car. A couple of years later, David Patterson, a computer scientist from the University of California Berkeley, and colleagues, estimated GPT-3’s carbon footprint at over 550,000 kg of CO2 In 2022, the tech company Hugging Face, estimated the carbon footprint of its own language model, BLOOM, as 25,000 kg in CO2 emissions. (BLOOM’s footprint is lower because Hugging Face uses renewable energy, but it doubled when other life-cycle processes like hardware manufacturing and use were added.)
Luckily, despite the growing size and numbers of data centers, their increasing energy demands and emissions have not kept pace proportionately—thanks to renewable energy sources and energy-efficient hardware.
But emissions don’t tell the full story.
AI’s hidden human cost
“If historical colonialism annexed territories, their resources, and the bodies that worked on them, data colonialism’s power grab is both simpler and deeper: the capture and control of human life itself through appropriating the data that can be extracted from it for profit.” So write Nick Couldry and Ulises Mejias, authors of the book The Costs of Connection.
The energy requirements, hardware manufacture and the cheap human labor behind AI systems disproportionately affect marginalized communities.
Technologies we use daily inexorably gather our data. “Human experience, potentially every layer and aspect of it, is becoming the target of profitable extraction,” Couldry and Meijas say. This feeds data capitalism, the economic model built on the extraction and commodification of data. While we are being dispossessed of our data, Big Tech commodifies it for their own benefit. This results in consolidation of power structures that reinforce existing race, gender, class and other inequalities.
“The political economy around tech and tech companies, and the development in advances in AI contribute to massive displacement and pollution, and significantly changes the built environment,” says technologist and activist Yeshi Milner, who founded Data For Black Lives (D4BL) to create measurable change in Black people’s lives using data. The energy requirements, hardware manufacture and the cheap human labor behind AI systems disproportionately affect marginalized communities.
AI’s recent explosive growth spiked the demand for manual, behind-the-scenes tasks, creating an industry described by Mary Gray and Siddharth Suri as “ghost work” in their book. This invisible human workforce that lies behind the “magic” of AI, is overworked and underpaid, and very often based in the Global South. For example, workers in Kenya who made less than $2 an hour, were the behind the mechanism that trained ChatGPT to properly talk about violence, hate speech and sexual abuse. And, according to an article in Analytics India Magazine, in some cases these workers may not have been paid at all, a case for wage theft. An exposé by the Washington Post describes “digital sweatshops” in the Philippines, where thousands of workers experience low wages, delays in payment, and wage theft by Remotasks, a platform owned by Scale AI, a $7 billion dollar American startup. Rights groups and labor researchers have flagged Scale AI as one company that flouts basic labor standards for workers abroad.
It is possible to draw a parallel with chattel slavery—the most significant economic event that continues to shape the modern world—to see the business structures that allow for the massive exploitation of people, Milner says. Back then, people got chocolate, sugar, cotton; today, they get generative AI tools. “What’s invisible through distance—because [tech companies] also control what we see—is the massive exploitation,” Milner says.
“At Data for Black Lives, we are less concerned with whether AI will become human…[W]e’re more concerned with the growing power of AI to decide who’s human and who’s not,” Milner says. As a decision-making force, AI becomes a “justifying factor for policies, practices, rules that not just reinforce, but are currently turning the clock back generations years on people’s civil and human rights.”
Ironically, AI plays an important role in mitigating its own harms—by plowing through mountains of data about weather changes, extreme weather events and human displacement.
Nuria Oliver, a computer scientist, and co-founder and vice-president of the European Laboratory of Learning and Intelligent Systems (ELLIS), says that instead of focusing on the hypothetical existential risks of today’s AI, we should talk about its real, tangible risks.
“Because AI is a transverse discipline that you can apply to any field [from education, journalism, medicine, to transportation and energy], it has a transformative power…and an exponential impact,” she says.
AI's accountability
“At the core of what we were arguing about data capitalism [is] a call to action to abolish Big Data,” says Milner. “Not to abolish data itself, but the power structures that concentrate [its] power in the hands of very few actors.”
A comprehensive AI Act currently negotiated in the European Parliament aims to rein Big Tech in. It plans to introduce a rating of AI tools based on the harms caused to humans, while being as technology-neutral as possible. That sets standards for safe, transparent, traceable, non-discriminatory, and environmentally friendly AI systems, overseen by people, not automation. The regulations also ask for transparency in the content used to train generative AIs, particularly with copyrighted data, and also disclosing that the content is AI-generated. “This European regulation is setting the example for other regions and countries in the world,” Oliver says. But, she adds, such transparencies are hard to achieve.
Google, for example, recently updated its privacy policy to say that anything on the public internet will be used as training data. “Obviously, technology companies have to respond to their economic interests, so their decisions are not necessarily going to be the best for society and for the environment,” Oliver says. “And that’s why we need strong research institutions and civil society institutions to push for actions.” ELLIS also advocates for data centers to be built in locations where the energy can be produced sustainably.
Ironically, AI plays an important role in mitigating its own harms—by plowing through mountains of data about weather changes, extreme weather events and human displacement. “The only way to make sense of this data is using machine learning methods,” Oliver says.
Milner believes that the best way to expose AI-caused systemic inequalities is through people's stories. “In these last five years, so much of our work [at D4BL] has been creating new datasets, new data tools, bringing the data to life. To show the harms but also to continue to reclaim it as a tool for social change and for political change.” This change, she adds, will depend on whose hands it is in.
Ever since he was a baby, Sharon Wong’s son Brandon suffered from rashes, prolonged respiratory issues and vomiting. In 2006, as a young child, he was diagnosed with a severe peanut allergy.
"My son had a history of reacting to traces of peanuts in the air or in food,” says Wong, a food allergy advocate who runs a blog focusing on nut free recipes, cooking techniques and food allergy awareness. “Any participation in school activities, social events, or travel with his peanut allergy required a lot of preparation.”
Peanut allergies affect around a million children in the U.S. Most never outgrow the condition. The problem occurs when the immune system mistakenly views the proteins in peanuts as a threat and releases chemicals to counteract it. This can lead to digestive problems, hives and shortness of breath. For some, like Wong’s son, even exposure to trace amounts of peanuts could be life threatening. They go into anaphylactic shock and need to take a shot of adrenaline as soon as possible.
Typically, people with peanut allergies try to completely avoid them and carry an adrenaline autoinjector like an EpiPen in case of emergencies. This constant vigilance is very stressful, particularly for parents with young children.
“The search for a peanut allergy ‘cure’ has been a vigorous one,” says Claudia Gray, a pediatrician and allergist at Vincent Pallotti Hospital in Cape Town, South Africa. The closest thing to a solution so far, she says, is the process of desensitization, which exposes the patient to gradually increasing doses of peanut allergen to build up a tolerance. The most common type of desensitization is oral immunotherapy, where patients ingest small quantities of peanut powder. It has been effective but there is a risk of anaphylaxis since it involves swallowing the allergen.
"By the end of the trial, my son tolerated approximately 1.5 peanuts," Sharon Wong says.
DBV Technologies, a company based in Montrouge, France has created a skin patch to address this problem. The Viaskin Patch contains a much lower amount of peanut allergen than oral immunotherapy and delivers it through the skin to slowly increase tolerance. This decreases the risk of anaphylaxis.
Wong heard about the peanut patch and wanted her son to take part in an early phase 2 trial for 4-to-11-year-olds.
“We felt that participating in DBV’s peanut patch trial would give him the best chance at desensitization or at least increase his tolerance from a speck of peanut to a peanut,” Wong says. “The daily routine was quite simple, remove the old patch and then apply a new one. By the end of the trial, he tolerated approximately 1.5 peanuts.”
How it works
For DBV Technologies, it all began when pediatric gastroenterologist Pierre-Henri Benhamou teamed up with fellow professor of gastroenterology Christopher Dupont and his brother, engineer Bertrand Dupont. Together they created a more effective skin patch to detect when babies have allergies to cow's milk. Then they realized that the patch could actually be used to treat allergies by promoting tolerance. They decided to focus on peanut allergies first as the more dangerous.
The Viaskin patch utilizes the fact that the skin can promote tolerance to external stimuli. The skin is the body’s first defense. Controlling the extent of the immune response is crucial for the skin. So it has defense mechanisms against external stimuli and can promote tolerance.
The patch consists of an adhesive foam ring with a plastic film on top. A small amount of peanut protein is placed in the center. The adhesive ring is attached to the back of the patient's body. The peanut protein sits above the skin but does not directly touch it. As the patient sweats, water droplets on the inside of the film dissolve the peanut protein, which is then absorbed into the skin.
The peanut protein is then captured by skin cells called Langerhans cells. They play an important role in getting the immune system to tolerate certain external stimuli. Langerhans cells take the peanut protein to lymph nodes which activate T regulatory cells. T regulatory cells suppress the allergic response.
A different patch is applied to the skin every day to increase tolerance. It’s both easy to use and convenient.
“The DBV approach uses much smaller amounts than oral immunotherapy and works through the skin significantly reducing the risk of allergic reactions,” says Edwin H. Kim, the division chief of Pediatric Allergy and Immunology at the University of North Carolina, U.S., and one of the principal investigators of Viaskin’s clinical trials. “By not going through the mouth, the patch also avoids the taste and texture issues. Finally, the ability to apply a patch and immediately go about your day may be very attractive to very busy patients and families.”
Brandon Wong displaying origami figures he folded at an Origami Convention in 2022
Sharon Wong
Clinical trials
Results from DBV's phase 3 trial in children ages 1 to 3 show its potential. For a positive result, patients who could not tolerate 10 milligrams or less of peanut protein had to be able to manage 300 mg or more after 12 months. Toddlers who could already tolerate more than 10 mg needed to be able to manage 1000 mg or more. In the end, 67 percent of subjects using the Viaskin patch met the target as compared to 33 percent of patients taking the placebo dose.
“The Viaskin peanut patch has been studied in several clinical trials to date with promising results,” says Suzanne M. Barshow, assistant professor of medicine in allergy and asthma research at Stanford University School of Medicine in the U.S. “The data shows that it is safe and well-tolerated. Compared to oral immunotherapy, treatment with the patch results in fewer side effects but appears to be less effective in achieving desensitization.”
The primary reason the patch is less potent is that oral immunotherapy uses a larger amount of the allergen. Additionally, absorption of the peanut protein into the skin could be erratic.
Gray also highlights that there is some tradeoff between risk and efficacy.
“The peanut patch is an exciting advance but not as effective as the oral route,” Gray says. “For those patients who are very sensitive to orally ingested peanut in oral immunotherapy or have an aversion to oral peanut, it has a use. So, essentially, the form of immunotherapy will have to be tailored to each patient.” Having different forms such as the Viaskin patch which is applied to the skin or pills that patients can swallow or dissolve under the tongue is helpful.
The hope is that the patch’s efficacy will increase over time. The team is currently running a follow-up trial, where the same patients continue using the patch.
“It is a very important study to show whether the benefit achieved after 12 months on the patch stays stable or hopefully continues to grow with longer duration,” says Kim, who is an investigator in this follow-up trial.
"My son now attends university in Massachusetts, lives on-campus, and eats dorm food. He has so much more freedom," Wong says.
The team is further ahead in the phase 3 follow-up trial for 4-to-11-year-olds. The initial phase 3 trial was not as successful as the trial for kids between one and three. The patch enabled patients to tolerate more peanuts but there was not a significant enough difference compared to the placebo group to be definitive. The follow-up trial showed greater potency. It suggests that the longer patients are on the patch, the stronger its effects.
They’re also testing if making the patch bigger, changing the shape and extending the minimum time it’s worn can improve its benefits in a trial for a new group of 4-to-11 year-olds.
The future
DBV Technologies is using the skin patch to treat cow’s milk allergies in children ages 1 to 17. They’re currently in phase 2 trials.
As for the peanut allergy trials in toddlers, the hope is to see more efficacy soon.
For Wong’s son who took part in the earlier phase 2 trial for 4-to-11-year-olds, the patch has transformed his life.
“My son continues to maintain his peanut tolerance and is not affected by peanut dust in the air or cross-contact,” Wong says. ”He attends university in Massachusetts, lives on-campus, and eats dorm food. He still carries an EpiPen but has so much more freedom than before his clinical trial. We will always be grateful.”
Scientists aim to preserve donkeys, one frozen embryo at a time
Every day for a week in 2022, Andres Gambini, a veterinarian and senior lecturer in animal science at the University of Queensland in Australia, walked into his lab—and headed straight to the video camera. Trained on an array of about 50 donkey embryos, all created by Gambini’s manual in vitro fertilization, or IVF, the camera kept an eye on their developmental progress. To eventually create a viable embryo that could be implanted into a female donkey, the embryos’ cells had to keep dividing, first in two, then in four and so on.
But the embryos weren’t cooperating. Some would start splitting up only to stop a day or two later, and others wouldn’t start at all. Every day he came in, Gambini saw fewer and fewer dividing embryos, so he was losing faith in the effort. “You see many failed attempts and get disappointed,” he says.
Gambini and his team, a group of Argentinian and Spanish researchers, were working to create these embryos because many donkey populations around the world are declining. It may sound counterintuitive that domesticated animals may need preservation, but out of 28 European donkey breeds, 20 are endangered and seven are in critical status. It is partly because of the inbreeding that happened over the course of many years and partly because in today’s Western world donkeys aren’t really used anymore.
“That's the reason why some breeds begin to disappear because humans were not really interested in having that specific breed anymore,” Gambini says. Nonetheless, in Africa, India and Latin America millions of rural families still rely on these hardy creatures for agriculture and transportation. And the only two wild donkey species—Equus africanus in Africa and Equus hemionus in Asia—are also dwindling, due to losing their habitats to human activities, diseases and slow reproduction rates. Gambini’s team wanted to create a way to preserve the animals for the future. “Donkeys are more endangered than people realize,” he says.
There’s much more to donkeys' trouble though. For the past 20 or so years, they have been facing a huge existential threat due to their hide gelatin, a compound derived from their skins by soaking and stewing. In Chinese traditional medicine, the compound, called ejiao, is believed to have a medicinal value, so it’s used in skin creams, added to food and taken in capsules. Centuries ago, ejiao was a very expensive luxury product available only for the emperor and his household. That changed in the 1990s when the Chinese economy boomed, and many people were suddenly able to afford it. “It went from a very elite product to a very popular product,” says Janneke Merkx, a campaign manager at The Donkey Sanctuary, a United Kingdom-based nonprofit organization that keeps tabs on the animals’ welfare worldwide. “It is a status symbol for gift giving.”
Having evolved in the harsh and arid mountainous terrains where food and water were scarce, donkeys are extremely adaptable and hardy. But the Donkey Sanctuary documented cases in which an entire village had their animals disappear overnight, finding them killed and skinned outside their settlement.
The Chinese donkey population was quickly decimated. Unlike many other farm animals, donkeys are finicky breeders. When stressed and unhappy, they don’t procreate, so growing them in large industrial settings isn’t possible. “Donkeys are notoriously slow breeders and really very difficult to farm,” says Merkx. “They are not the same as other livestock like sheep and pigs and cattle.” Within years the, the donkey numbers in China dropped precipitously. “China used to have the largest donkey population in the world in the 1990s. They had 11 million donkeys, and it's now down to less than 3 million, and they just can't keep up with the demand.”
To keep the ejiao conveyor going, some producers turned to the illegal wildlife trade. Poachers began to steal and slaughter donkeys from rural villages in Africa. The Donkey Sanctuary documented cases in which an entire village had their animals disappear overnight, finding them killed and skinned outside their settlement. Exactly how many creatures were lost to the skin trade to-date isn’t possible to calculate, says Faith Burden, the Donkey Sanctuary’s director of equine operations. Traditionally a poor people’s beast of burden, donkey counts are hard to keep track of. “When an animal doesn't produce meat, milk or eggs or whatever edible product, they're often less likely to be acknowledged in a government population census,” Burden says. “So reliable statistics are hard to come by.” The nonprofit estimates that about 4.8 million are slaughtered annually.
During their six to seven thousand years of domestication, donkeys rarely got the full appreciation for their services. They are often compared to horses, which doesn’t do them justice. They’re entirely different animals, Burden says. Built for speed, horses respond to predators and other dangers by running as fast as they can. Donkeys, which originate from the rocky, mountainous regions of Africa where running is dangerous, react to threats by freezing and assessing the situation for the best response. “Those so-called stubborn donkeys that won’t move as you want, they are actually thinking ‘what’s the best approach,’” Burden says. They may even choose to fight the predators rather than flee, she adds. “In some parts of the world, people use them as guard animals against things like coyotes and wolves.”
Scientists believe that domestic donkeys take their origin from Equus africanus or African wild ass, originally roaming where Kenya, Ethiopia and Eritrea are today. Having evolved in the harsh and arid mountainous terrains where food and water were scarce, they are extremely adaptable and hardy. Research finds that they can go without water for 72 hours and then drink their fill without any negative consequences. Their big jaws let them chew tough desert shrubs, which horses can’t exist on. Their large ears help dissipate heat. Their little upright hooves are a perfect fit for the uneven rocky or other dangerous grounds. Accustomed to the mountain desert climate with hot days and cold nights, they don’t mind temperature flux.
“The donkey is the most supremely adapted animal to deal with hostile conditions,” Burden says. “They can survive on much lower nutritional quality food than a cow, sheep or horse. That’s why communities living in some of the most inhospitable places will often have donkeys with them.” And that’s why losing a donkey to an illegal skin trade can devastate a family in places like Eritrea. Suddenly everything from water to firewood to produce must be carried by family members—and often women.
Workers unloading donkeys at the Shinyanga slaughterhouse in Tanzania. Fearing a future in which donkeys go extinct, scientists have found ways to cryopreserve a donkey embryo in liquid nitrogen.
TAHUCHA
One can imagine a time when worldwide donkey populations may dwindle to the point that they would need to be restored. That includes their genetic variability too. That’s where the frozen embryos may come in handy. We may be able to use them to increase the genetic variability of donkeys, which will be especially important if they get closer to extinction, Gambini says. His team had already created frozen embryos for horses and zebras, an idea similar to a seed bank. “We call this concept the Frozen Zoo.”
Creating donkey embryos proved much harder than those of zebras and horses. To improve chances of fertilization, Gambini used the intracytoplasmic sperm injection or ICSI, in which he employed a tiny needle called a micropipette to inject a donkey sperm into an egg. That was a step above the traditional IVF method, in which the egg and a sperm are left floating in a test tube together. The injection took, but during the incubating week, one after the other, the embryos stopped dividing. Finally, on day seven, Gambini finally spotted the exact sight he was hoping to see. One of the embryos developed into a burgeoning ball of cells.
“That stage is called a blastocyst,” Gambini says. The clump of cells had a lot of fluids mixed within them, which indicated that they were finally developing into a viable embryo. “When we see a blastocyst, we know we can transfer that into a female.” He was so excited he immediately called all his collaborators to tell them the good news, which they later published in the journal of Theriogenology.
The one and only embryo to reach that stage, the blastocyst was cryopreserved in liquid nitrogen. The team is waiting for the next breeding season to see if a female donkey may carry it to term and give birth to a healthy foal. Gambini’s team is hoping to polish the process and create more embryos. “It’s our weapon in the conservation ass-enal,” he says.
Lina Zeldovich has written about science, medicine and technology for Popular Science, Smithsonian, National Geographic, Scientific American, Reader’s Digest, the New York Times and other major national and international publications. A Columbia J-School alumna, she has won several awards for her stories, including the ASJA Crisis Coverage Award for Covid reporting, and has been a contributing editor at Nautilus Magazine. In 2021, Zeldovich released her first book, The Other Dark Matter, published by the University of Chicago Press, about the science and business of turning waste into wealth and health. You can find her on http://linazeldovich.com/ and @linazeldovich.