Massive benefits of AI come with environmental and human costs. Can AI itself be part of the solution?
Generative AI has a large carbon footprint and other drawbacks. But AI can help mitigate its own harms—by plowing through mountains of data on extreme weather and human displacement.
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.
Joy Milne's unusual sense of smell led Dr. Tilo Kunath, a neurobiologist at the Centre for Regenerative Medicine at the University of Edinburgh, and a host of other scientists, to develop a new diagnostic test for Parkinson's.
Forty years ago, Joy Milne, a nurse from Perth, Scotland, noticed a musky odor coming from her husband, Les. At first, Milne thought the smell was a result of bad hygiene and badgered her husband to take longer showers. But when the smell persisted, Milne learned to live with it, not wanting to hurt her husband's feelings.
Twelve years after she first noticed the "woodsy" smell, Les was diagnosed at the age of 44 with Parkinson's Disease, a neurodegenerative condition characterized by lack of dopamine production and loss of movement. Parkinson's Disease currently affects more than 10 million people worldwide.
Milne spent the next several years believing the strange smell was exclusive to her husband. But to her surprise, at a local support group meeting in 2012, she caught the familiar scent once again, hanging over the group like a cloud. Stunned, Milne started to wonder if the smell was the result of Parkinson's Disease itself.
Milne's discovery led her to Dr. Tilo Kunath, a neurobiologist at the Centre for Regenerative Medicine at the University of Edinburgh. Together, Milne, Kunath, and a host of other scientists would use Milne's unusual sense of smell to develop a new diagnostic test, now in development and poised to revolutionize the treatment of Parkinson's Disease.
"Joy was in the audience during a talk I was giving on my work, which has to do with Parkinson's and stem cell biology," Kunath says. "During the patient engagement portion of the talk, she asked me if Parkinson's had a smell to it." Confused, Kunath said he had never heard of this – but for months after his talk he continued to turn the question over in his mind.
Kunath knew from his research that the skin's microbiome changes during different disease processes, releasing metabolites that can give off odors. In the medical literature, diseases like melanoma and Type 2 diabetes have been known to carry a specific scent – but no such connection had been made with Parkinson's. If people could smell Parkinson's, he thought, then it stood to reason that those metabolites could be isolated, identified, and used to potentially diagnose Parkinson's by their presence alone.
First, Kunath and his colleagues decided to test Milne's sense of smell. "I got in touch with Joy again and we designed a protocol to test her sense of smell without her having to be around patients," says Kunath, which could have affected the validity of the test. In his spare time, Kunath collected t-shirt samples from people diagnosed with Parkinson's and from others without the diagnosis and gave them to Milne to smell. In 100 percent of the samples, Milne was able to detect whether a person had Parkinson's based on smell alone. Amazingly, Milne was even able to detect the "Parkinson's scent" in a shirt from the control group – someone who did not have a Parkinson's diagnosis, but would go on to be diagnosed nine months later.
From the initial study, the team discovered that Parkinson's did have a smell, that Milne – inexplicably – could detect it, and that she could detect it long before diagnosis like she had with her husband, Les. But the experiments revealed other things that the team hadn't been expecting.
"One surprising thing we learned from that experiment was that the odor was always located in the back of the shirt – never in the armpit, where we expected the smell to be," Kunath says. "I had a chance meeting with a dermatologist and he said the smell was due to the patient's sebum, which are greasy secretions that are really dense on your upper back. We have sweat glands, instead of sebum, in our armpits." Patients with Parkinson's are also known to have increased sebum production.
With the knowledge that a patient's sebum was the source of the unusual smell, researchers could go on to investigate exactly what metabolites were in the sebum and in what amounts. Kunath, along with his associate, Dr. Perdita Barran, collected and analyzed sebum samples from 64 participants across the United Kingdom. Once the samples were collected, Barran and others analyzed it using a method called gas chromatography mass spectrometry, or GS-MC, which separated, weighed and helped identify the individual compounds present in each sebum sample.
Barran's team can now correctly identify Parkinson's in nine out of 10 patients – a much quicker and more accurate way to diagnose than what clinicians do now.
"The compounds we've identified in the sebum are not unique to people with Parkinson's, but they are differently expressed," says Barran, a professor of mass spectrometry at the University of Manchester. "So this test we're developing now is not a black-and-white, do-you-have-something kind of test, but rather how much of these compounds do you have compared to other people and other compounds." The team identified over a dozen compounds that were present in the sebum of Parkinson's patients in much larger amounts than the control group.
Using only the GC-MS and a sebum swab test, Barran's team can now correctly identify Parkinson's in nine out of 10 patients – a much quicker and more accurate way to diagnose than what clinicians do now.
"At the moment, a clinical diagnosis is based on the patient's physical symptoms," Barran says, and determining whether a patient has Parkinson's is often a long and drawn-out process of elimination. "Doctors might say that a group of symptoms looks like Parkinson's, but there are other reasons people might have those symptoms, and it might take another year before they're certain," Barran says. "Some of those symptoms are just signs of aging, and other symptoms like tremor are present in recovering alcoholics or people with other kinds of dementia." People under the age of 40 with Parkinson's symptoms, who present with stiff arms, are often misdiagnosed with carpal tunnel syndrome, she adds.
Additionally, by the time physical symptoms are present, Parkinson's patients have already lost a substantial amount of dopamine receptors – about sixty percent -- in the brain's basal ganglia. Getting a diagnosis before physical symptoms appear would mean earlier interventions that could prevent dopamine loss and preserve regular movement, Barran says.
"Early diagnosis is good if it means there's a chance of early intervention," says Barran. "It stops the process of dopamine loss, which means that motor symptoms potentially will not happen, or the onset of symptoms will be substantially delayed." Barran's team is in the processing of streamlining the sebum test so that definitive results will be ready in just two minutes.
"What we're doing right now will be a very inexpensive test, a rapid-screen test, and that will encourage people to self-sample and test at home," says Barran. In addition to diagnosing Parkinson's, she says, this test could also be potentially useful to determine if medications were at a therapeutic dose in people who have the disease, since the odor is strongest in people whose symptoms are least controlled by medication.
"When symptoms are under control, the odor is lower," Barran says. "Potentially this would allow patients and clinicians to see whether their symptoms are being managed properly with medication, or perhaps if they're being overmedicated." Hypothetically, patients could also use the test to determine if interventions like diet and exercise are effective at keeping Parkinson's controlled.
"We hope within the next two to five years we will have a test available."
Barran is now running another clinical trial – one that determines whether they can diagnose at an earlier stage and whether they can identify a difference in sebum samples between different forms of Parkinson's or diseases that have Parkinson's-like symptoms, such as Lewy Body Dementia.
"Within the next one to two years, we hope to be running a trial in the Manchester area for those people who do not have motor symptoms but are at risk for developing dementia due to symptoms like loss of smell and sleep difficulty," Barran had said in 2019. "If we can establish that, we can roll out a test that determines if you have Parkinson's or not with those first pre-motor symptoms, and then at what stage. We hope within the next two to five years we will have a test available."
In a 2022 study, published in the American Chemical Society, researchers used mass spectrometry to analyze sebum from skin swabs for the presence of the specific molecules. They found that some specific molecules are present only in people who have Parkinson’s. Now they hope that the same method can be used in regular diagnostic labs. The test, many years in the making, is inching its way to the clinic.
"We would likely first give this test to people who are at risk due to a genetic predisposition, or who are at risk based on prodomal symptoms, like people who suffer from a REM sleep disorder who have a 50 to 70 percent chance of developing Parkinson's within a ten year period," Barran says. "Those would be people who would benefit from early therapeutic intervention. For the normal population, it isn't beneficial at the moment to know until we have therapeutic interventions that can be useful."
Milne's husband, Les, passed away from complications of Parkinson's Disease in 2015. But thanks to him and the dedication of his wife, Joy, science may have found a way to someday prolong the lives of others with this devastating disease. Sometimes she can smell people who have Parkinson’s while in the supermarket or walking down the street but has been told by medical ethicists she cannot tell them, Milne said in an interview with the Guardian. But once the test becomes available in the clinics, it will do the job for her.
[Ed. Note: A older version of this hit article originally ran on September 3, 2019.]
Chris Mirabile sprints on a track in Sarasota, Florida, during his daily morning workout. He claims to be a superager already, at age 38, with test results to back it up.
Earlier this year, Harvard scientists reported that they used an anti-aging therapy to reverse blindness in elderly mice. Several other studies in the past decade have suggested that the aging process can be modified, at least in lab organisms. Considering mice and humans share virtually the same genetic makeup, what does the rodent-based study mean for the humans?
In truth, we don’t know. Maybe nothing.
What we do know, however, is that a growing number of people are dedicating themselves to defying the aging process, to turning back the clock – the biological clock, that is. Take Bryan Johnson, a man who is less mouse than human guinea pig. A very wealthy guinea pig.
The 45-year-old venture capitalist spends over $2 million per year reversing his biological clock. To do this, he employs a team of 30 medical doctors and other scientists. His goal is to eventually reset his biological clock to age 18, and “have all of his major organs — including his brain, liver, kidneys, teeth, skin, hair, penis and rectum — functioning as they were in his late teens,” according to a story earlier this year in the New York Post.
But his daily routine paints a picture that is far from appealing: for example, rigorously adhering to a sleep schedule of 8 p.m. to 5 a.m. and consuming more than 100 pills and precisely 1,977 calories daily. Considering all of Johnson’s sacrifices, one discovers a paradox:
To live forever, he must die a little every day until he reaches his goal - if he ever reaches his goal.
Less extreme examples seem more helpful for people interested in happy, healthy aging. Enter Chris Mirabile, a New Yorker who says on his website, SlowMyAge.com, that he successfully reversed his biological age by 13.6 years, from the chronological age of 37.2 to a biological age of 23.6. To put this achievement in perspective, Johnson, to date, has reversed his biological clock by 2.5 years.
Mirabile's habits and overall quest to turn back the clock trace back to a harrowing experience at age 16 during a school trip to Manhattan, when he woke up on the floor with his shirt soaked in blood.
Mirabile, who is now 38, supports his claim with blood tests that purport to measure biological age by assessing changes to a person’s epigenome, or the chemical marks that affect how genes are expressed. Mirabile’s tests have been run and verified independently by the same scientific lab that analyzes Johnson’s. (In an email to Leaps.org, the lab, TruDiagnostic, confirmed Mirabile’s claims about his test results.)
There is considerable uncertainty among scientists about the extent to which these tests can accurately measure biological age in individuals. Even so, Mirabile’s results are intriguing. They could reflect his smart lifestyle for healthy aging.
His habits and overall quest to turn back the clock trace back to a harrowing experience at age 16 during a school trip to Manhattan, when Mirabile woke up on the floor with his shirt soaked in blood. He’d severed his tongue after a seizure. He later learned it was caused by a tumor the size of a golf ball. As a result, “I found myself contemplating my life, what I had yet to experience, and mortality – a theme that stuck with me during my year of recovery and beyond,” Mirabile told me.
For the next 15 years, he researched health and biology, integrating his learnings into his lifestyle. Then, in his early 30s, he came across an article in the journal Cell, "The Hallmarks of Aging," that outlined nine mechanisms of the body that define the aging process. Although the paper says there are no known interventions to delay some of these mechanisms, others, such as inflammation, struck Mirabile as actionable. Reading the paper was his “moment of epiphany” when it came to the areas where he could assert control to maximize his longevity.
He also wanted “to create a resource that my family, friends, and community could benefit from in the short term,” he said. He turned this knowledge base into a company called NOVOS dedicated to extending lifespan.
His longevity advice is more accessible than Johnson’s multi-million dollar approach, as Mirabile spends a fraction of that amount. Mirabile takes one epigenetic test per year and has a gym membership at $45 per month. Unlike Johnson, who takes 100 pills per day, Mirabile takes 10, costing another $45 monthly, including a B-complex, fish oil, Vitamins D3 and K2, and two different multivitamin supplements.
Mirabile’s methods may be easier to apply in other ways as well, since they include activities that many people enjoy anyway. He’s passionate about outdoor activities, travels frequently, and has loving relationships with friends and family, including his girlfriend and collie.
Here are a few of daily routines that could, he thinks, contribute to his impressively young bio age:
After waking at 7:45 am, he immediately drinks 16 ounces of water, with 1/4 teaspoon of sodium and potassium to replenish electrolytes. He takes his morning vitamins, brushes and flosses his teeth, puts on a facial moisturizing sunblock and goes for a brisk, two-mile walk in the sun. At 8:30 am on Mondays, Wednesdays, and Fridays he lift weights, focusing on strength and power, especially in large muscle groups.
Tuesdays, Thursdays and Saturdays are intense cardio days. He runs 5-7 miles or bicycles for 60 minutes first thing in the morning at a brisk pace, listening to podcasts. Sunday morning cardio is more leisurely.
After working out each day, he’s back home at 9:20 am, where he makes black coffee, showers, then applies serum and moisturizing sunblock to his face. He works for about three hours on his laptop, then has a protein shake and fruit.
Mirabile is a dedicated intermittent faster, with a six hour eating window in between 18 hours fasts. At 3 pm, he has lunch. The Mediterranean lineup often features salmon, sardines, olive oil, pink Himalayan salt plus potassium salt for balance, and lots of dried herbs and spices. He almost always finishes with 1/3 to 1/2 bar of dark chocolate.
If you are what you eat, Mirabile is made of mostly plants and lean meats. He follows a Mediterranean diet full of vegetables, fruits, fatty fish and other meats full of protein and unsaturated fats. “These may cost more than a meal at an American fast-food joint, but then again, not by much,” he said. Each day, he spends $25 on all his meals combined.
At 6 pm, he takes the dog out for a two-mile walk, taking calls for work or from family members along the way. At 7 pm, he dines with his girlfriend. Like lunch, this meal is heavy on widely available ingredients, including fish, fresh garlic, and fermented food like kimchi. Mirabile finishes this meal with sweets, like coconut milk yogurt with cinnamon and clove, some stevia, a mix of fresh berries and cacao nibs.
If Mirabile's epigenetic tests are accurate, his young biological age could be thanks to his healthy lifestyle, or it could come from a stroke of luck if he inherited genes that protect against aging.
At 8 pm, he wraps up work duties and watches shows with his girlfriend, applies serum and moisturizer yet again, and then meditates with the lights off. This wind-down, he said, improves his sleep quality. Wearing a sleep mask and earplugs, he’s asleep by about 10:30.
“I’ve achieved stellar health outcomes, even after having had the physiological stressors of a brain tumor, without spending a fortune,” Mirabile said. “In fact, even during times when I wasn’t making much money as a startup founder with few savings, I still managed to live a very healthy, pro-longevity lifestyle on a modest budget.”
Mirabile said living a cleaner, healthier existence is a reality that many readers can achieve. It’s certainly true that many people live in food deserts and have limited time for exercise or no access to gyms, but James R. Doty, a clinical professor of neurosurgery at Stanford, thinks many can take more action to stack the odds that they’ll “be happy and live longer.” Many of his recommendations echo aspects of Mirabile’s lifestyle.
Each night, Doty said, it’s vital to get anywhere between 6-8 hours of good quality sleep. Those who sleep less than 6 hours per night are at an increased risk of developing a whole host of medical problems, including high blood pressure, type 2 diabetes, and stroke.
In addition, it’s critical to follow Mirabile’s prescription of exercise for about one hour each day, and intensity levels matter. Doty noted that, in 2017, researchers at Brigham Young University found that people who ran at a fast pace for 30-40 minutes five days per week were, on average, biologically younger by nine years, compared to those who subscribed to more moderate exercise programs, as well as those who rarely exercised.
When it comes to nutrition, one should consider fasting for 16 hours per day, Doty said. This is known as the 16/8 method, where one’s daily calories are consumed within an eight hour window, fasting for the remaining 16 hours, just like Mirabile. Intermittent fasting is associated with cellular repair and less inflammation, though it’s not for everyone, Doty added. Consult with a medical professional before trying a fasting regimen.
Finally, Doty advised to “avoid anger, avoid stress.” Easier said than done, but not impossible. “Between stimulus and response, there is a pause and within that pause lies your freedom,” Doty said. Mirabile’s daily meditation ritual could be key to lower stress for healthy aging. Research has linked regular, long-term meditation to having a lower epigenetic age, compared to control groups.
Many other factors could apply. Having a life purpose, as Mirabile does with his company, has also been associated with healthy aging and lower epigenetic age. Of course, Mirabile is just one person, so it’s hard to know how his experience will apply to others. If his tests are accurate, his young biological age could be thanks to his healthy lifestyle, or it could come from a stroke of luck if he inherited genes that protect against aging. Clearly, though, any such genes did not protect him from cancer at an early age.
The third and perhaps most likely explanation: Mirabile’s very young biological age results from a combination of these factors. Some research shows that genetics account for only 25 percent of longevity. That means environmental factors could be driving the other 75 percent, such as where you live, frequency of exercise, quality of nutrition and social support.
The middle-aged – even Brian Johnson – probably can’t ever be 18 again. But more modest goals are reasonable for many. Control what you can for a longer, healthier life.