Leo S. Lo, University of Virginia

Artificial intelligence systems are thirsty, consuming as much as 500 milliliters of water – a single-serving water bottle – for each short conversation a user has with the GPT-3 version of OpenAI’s ChatGPT system. They use roughly the same amount of water to draft a 100-word email message.

That figure includes the water used to cool the data center’s servers and the water consumed at the power plants generating the electricity to run them.

But the study that calculated those estimates also pointed out that AI systems’ water usage can vary widely, depending on where and when the computer answering the query is running.

To me, as an academic librarian and professor of education, understanding AI is not just about knowing how to write prompts. It also involves understanding the infrastructure, the trade-offs, and the civic choices that surround AI.

Many people assume AI is inherently harmful, especially given headlines calling out its vast energy and water footprint. Those effects are real, but they’re only part of the story.

When people move from seeing AI as simply a resource drain to understanding its actual footprint, where the effects come from, how they vary, and what can be done to reduce them, they are far better equipped to make choices that balance innovation with sustainability.

2 hidden streams

Behind every AI query are two streams of water use.

The first is on-site cooling of servers that generate enormous amounts of heat. This often uses evaporative cooling towers – giant misters that spray water over hot pipes or open basins. The evaporation carries away heat, but that water is removed from the local water supply, such as a river, a reservoir or an aquifer. Other cooling systems may use less water but more electricity.

The second stream is used by the power plants generating the electricity to power the data center. Coal, gas and nuclear plants use large volumes of water for steam cycles and cooling.

Hydropower also uses up significant amounts of water, which evaporates from reservoirs. Concentrated solar plants, which run more like traditional steam power stations, can be water-intensive if they rely on wet cooling.

By contrast, wind turbines and solar panels use almost no water once built, aside from occasional cleaning.

Climate and timing matter

Water use shifts dramatically with location. A data center in cool, humid Ireland can often rely on outside air or chillers and run for months with minimal water use. By contrast, a data center in Arizona in July may depend heavily on evaporative cooling. Hot, dry air makes that method highly effective, but it also consumes large volumes of water, since evaporation is the mechanism that removes heat.

Timing matters too. A University of Massachusetts Amherst study found that a data center might use only half as much water in winter as in summer. And at midday during a heat wave, cooling systems work overtime. At night, demand is lower.

Newer approaches offer promising alternatives. For instance, immersion cooling submerges servers in fluids that don’t conduct electricity, such as synthetic oils, reducing water evaporation almost entirely.

And a new design from Microsoft claims to use zero water for cooling, by circulating a special liquid through sealed pipes directly across computer chips. The liquid absorbs heat and then releases it through a closed-loop system without needing any evaporation. The data centers would still use some potable water for restrooms and other staff facilities, but cooling itself would no longer draw from local water supplies.

These solutions are not yet mainstream, however, mainly because of cost, maintenance complexity and the difficulty of converting existing data centers to new systems. Most operators rely on evaporative systems.

A simple skill you can use

The type of AI model being queried matters, too. That’s because of the different levels of complexity and the hardware and amount of processor power they require. Some models may use far more resources than others. For example, one study found that certain models can consume over 70 times more energy and water than ultra‑efficient ones.

You can estimate AI’s water footprint yourself in just three steps, with no advanced math required.

Step 1 – Look for credible research or official disclosures. Independent analyses estimate that a medium-length GPT-5 response, which is about 150 to 200 words of output, or roughly 200 to 300 tokens, uses about 19.3 watt-hours. A response of similar length from GPT-4o uses about 1.75 watt-hours.

Step 2 – Use a practical estimate for the amount of water per unit of electricity, combining the usage for cooling and for power.

Independent researchers and industry reports suggest that a reasonable range today is about 1.3 to 2.0 milliliters per watt-hour. The lower end reflects efficient facilities that use modern cooling and cleaner grids. The higher end represents more typical sites.

Step 3 – Now it’s time to put the pieces together. Take the energy number you found in Step 1 and multiply it by the water factor from Step 2. That gives you the water footprint of a single AI response.

Here’s the one-line formula you’ll need:

Energy per prompt (watt-hours) × Water factor (milliliters per watt-hour) = Water per prompt (in milliliters)

For a medium-length query to GPT-5, that calculation should use the figures of 19.3 watt-hours and 2 milliliters per watt-hour. 19.3 x 2 = 39 milliliters of water per response.

For a medium-length query to GPT-4o, the calculation is 1.75 watt-hours x 2 milliliters per watt-hour = 3.5 milliliters of water per response.

If you assume the data centers are more efficient, and use 1.3 milliliters per watt-hour, the numbers drop: about 25 milliliters for GPT-5 and 2.3 milliliters for GPT-4o.

A recent Google technical report said a median text prompt to its Gemini system uses just 0.24 watt-hours of electricity and about 0.26 milliliters of water – roughly the volume of five drops. However, the report does not say how long that prompt is, so it can’t be compared directly with GPT water usage.

Those different estimates – ranging from 0.26 milliliters to 39 milliliters – demonstrate how much the effects of efficiency, AI model and power-generation infrastructure all matter.

Comparisons can add context

To truly understand how much water these queries use, it can be helpful to compare them to other familiar water uses.

When multiplied by millions, AI queries’ water use adds up. OpenAI reports about 2.5 billion prompts per day. That figure includes queries to its GPT-4o, GPT-4 Turbo, GPT-3.5 and GPT-5 systems, with no public breakdown of how many queries are issued to each particular model.

Using independent estimates and Google’s official reporting gives a sense of the possible range:

• All Google Gemini median prompts: about 650,000 liters per day.
• All GPT 4o medium prompts: about 8.8 million liters per day.
• All GPT 5 medium prompts: about 97.5 million liters per day.

For comparison, Americans use about 34 billion liters per day watering residential lawns and gardens. One liter is about one-quarter of a gallon.

Generative AI does use water, but – at least for now – its daily totals are small compared with other common uses such as lawns, showers and laundry.

But its water demand is not fixed. Google’s disclosure shows what is possible when systems are optimized, with specialized chips, efficient cooling and smart workload management. Recycling water and locating data centers in cooler, wetter regions can help, too.

Transparency matters, as well: When companies release their data, the public, policymakers and researchers can see what is achievable and compare providers fairly.

Leo S. Lo, Dean of Libraries; Advisor to the Provost for AI Literacy; Professor of Education, University of Virginia

This article is republished from The Conversation under a Creative Commons license. Read the original article.

David Kroll, University of Colorado Anschutz Medical Campus

David Bregger had never heard of kratom before his son, Daniel, 33, died in Denver in 2021 from using what he thought was a natural and safe remedy for anxiety.

By his father’s account, Daniel didn’t know that the herbal product could kill him. The product listed no ingredients or safe-dosing information on the label. And it had no warning that it should not be combined with other sedating drugs, such as the over-the-counter antihistamine diphenhydramine, which is the active ingredient in Benadryl and other sleep aids.

As the fourth anniversary of Daniel’s death approaches, a recently enacted Colorado law aims to prevent other families from experiencing the heartbreak shared by the Bregger family. Colorado Senate Bill 25-072, known as the Daniel Bregger Act, addresses what the state legislature calls the deceptive trade practices around the sale of concentrated kratom products artificially enriched with a chemical called 7-OH.

7-OH, known as 7-hydroxymitragynine, has also garnered national attention. On July 29, 2025, the U.S. Food and Drug Administration issued a warning that products containing 7-OH are potent opioids that can pose significant health risks and even death.

As kratom and its constituents are studied in greater detail, the Centers for Disease Control and Prevention and university researchers have documented hundreds of deaths where kratom-derived chemicals were present in postmortem blood tests. But rarely is kratom deadly by itself. In a study of 551 kratom-related deaths in Florida, 93.5% involved other substances such as opioids like fentanyl.

I study pharmaceutical sciences, have taught for over 30 years about herbal supplements like kratom, and I’ve written about kratom’s effects and controversy.

Kratom – one name, many products

Kratom is a broad term used to describe products made from the leaves of a Southeast Asian tree known scientifically as Mitragyna speciosa. The Latin name derives from the shape of its leaves, which resemble a bishop’s miter, the ceremonial, pointed headdress worn by bishops and other church leaders.

Kratom is made from dried and powdered leaves that can be chewed or made into a tea. Used by rice field workers and farmers in Thailand to increase stamina and productivity, kratom initially alleviates fatigue with an effect like that of caffeine. In larger amounts, it imparts a sense of well-being similar to opioids.

In fact, mitragynine, which is found in small amounts in kratom, partially stimulates opioid receptors in the central nervous system. These are the same type of opioid receptors that trigger the effects of drugs such as morphine and oxycodone. They are also the same receptors that can slow or stop breathing when overstimulated.

In the body, the small amount of mitragynine in kratom powder is converted to 7-OH by liver enzymes, hence the opioid-like effects in the body. 7-OH can also be made in a lab and is used to increase the potency of certain kratom products, including the ones found in gas stations or liquor stores.

And therein lies the controversy over the risks and benefits of kratom.

Natural or lab made: All medicines have risks

Because kratom is a plant-derived product, it has fallen into a murky enforcement area. It is sold as an herbal supplement, normally by the kilogram from online retailers overseas.

In 2016, I wrote a series of articles for Forbes as the Drug Enforcement Administration proposed to list kratom constituents on the most restrictive Schedule 1 of the Controlled Substances Act. This classification is reserved for drugs the DEA determines to possess “no currently accepted medical use and a high potential for abuse,” such as heroin and LSD.

But readers countered the DEA’s stance and sent me more than 200 messages that primarily documented their use of kratom as an alternative to opioids for pain.

Others described how kratom assisted them in recovery from addiction to alcohol or opioids themselves. Similar stories also flooded the official comments requested by the DEA, and the public pressure presumably led the agency to drop its plan to regulate kratom as a controlled substance.

But not all of the stories pointed to kratom’s benefits. Instead, some people pointed out a major risk: becoming addicted to kratom itself. I learned it is a double-edged sword – remedy to some, recreational risk to others. A national survey of kratom users was consistent with my nonscientific sampling, showing more than half were using the supplement to relieve pain, stress, anxiety or a combination of these.

Natural leaf powder vs. artificially concentrated extracts

After the DEA dropped its 2016 plan to ban the leaf powder, marketers in the U.S. began isolating mitragynine and concentrating it into small bottles that could be taken like those energy shots of caffeine often sold in gas stations and convenience stores. This formula made it easier to ingest more kratom. Slowly, sellers learned they could make the more potent 7-OH from mitragynine and give their products an extra punch. And an extra dose of risk.

People who use kratom in the powder form describe taking 3 to 5 grams, the size of a generous tablespoon. They put the powder in capsules or made it into a tea several times a day to ward off pain, the craving for alcohol or the withdrawal symptoms from long-term prescription opioid use. Since this form of kratom does not contain very much mitragynine – it is only about 1% of the powdered leaf – overdosing on the powder alone does not typically happen.

That, along with pushback from consumers, is why the Food and Drug Administration is proposing to restrict only the availability of 7-OH and not mitragynine or kratom powder. The new Colorado law limits the concentration of kratom ingredients in products and restricts their sales and marketing to consumers over 21.

Even David Bregger supports this distinction. “I’m not anti-kratom, I’m pro-regulation. What I’m after is getting nothing but leaf product,” he told WPRI in Rhode Island last year while demonstrating at a conference of the education and advocacy trade group the American Kratom Association.

Such lobbying with the trade group last year led the American Kratom Association to concur that 7-OH should be regulated as a Schedule 1 controlled substance. The association acknowledges that such regulation is reasonable and based in science.

Benefits amid the ban

Despite the local and national debate over 7-OH, scientists are continuing to explore kratom compounds for their legitimate medical use.

A $3.5 million NIH grant is one of several that is increasing understanding of kratom as a source for new drugs.

Researchers have identified numerous other chemicals called alkaloids from kratom leaf specimens and commercial products. These researchers show that some types of kratom trees make unique chemicals, possibly opening the door to other painkillers. Researchers have also found that compounds from kratom, such as 7-OH, bind to opioid receptors in unique ways. The compounds seem to have an effect more toward pain management and away from potentially deadly suppression of breathing. Of course, this is when the compounds are used alone and not together with other sedating drugs.

Rather than contributing to the opioid crisis, researchers suspect that isolated and safely purified drugs made from kratom could be potential treatments for opioid addiction. In fact, some kratom chemicals such as mitragynine have multiple actions and could potentially replace both medication-assisted therapy, like buprenorphine, in treating opioid addiction and drugs like clonidine for opioid withdrawal symptoms.

Rigorous scientific study has led to this more reasonable juncture in the understanding of kratom and its sensible regulation. Sadly, we cannot bring back Daniel Bregger. But researchers can advance the potential for new and beneficial drugs while legislators help prevent such tragedies from befalling other families.

David Kroll, Professor of Natural Products Pharmacology & Toxicology, University of Colorado Anschutz Medical Campus

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Almut Winterstein, University of Florida and Sonja Rasmussen, Johns Hopkins University

A panel convened in July 2025 by the Food and Drug Administration sparked controversy by casting doubt about the safety of commonly used antidepressants during pregnancy. But it also raised the broader issue of how little is known about the safety of many medications used in pregnancy, considering the implications for both mother and child – and how understudied this topic is.

In the U.S., the average pregnant patient takes four prescription medications, and more than 9 in 10 patients take at least one. But most drugs lack conclusive evidence about their safety during pregnancy. About 1 in 5 women uses a medication during pregnancy that has some preliminary evidence that it could cause harm but for which conclusive studies are missing.

We are researchers in maternal and child health who evaluate the safety of medications during pregnancy. In our work, we identify medications that might raise the risk for birth defects or pregnancy loss and compare the safety of different treatments.

While progress has been slow, researchers and federal agencies have built monitoring systems, databases and tools to accelerate our understanding of medication safety. However, these efforts are now at risk due to ongoing cuts to medical research funding – and with them, so is the knowledge base for determining whether sticking with a therapy or discontinuing it offers the safest choice for both mother and child.

How pregnant women got sidelined

One big reason why so little is known about the effects of medications during pregnancy stretches back more than half a century. In the 1960s, a drug called thalidomide that was widely prescribed to treat morning sickness in pregnant women caused severe birth defects in over 10,000 children around the world. In response, in 1977 the FDA recommended excluding women of childbearing age from participating in early stage clinical trials testing new medications.

Ethically, there is long-standing tension between concerns about fetal harm and maternal needs. Legal liability and added complexities when conducting studies in pregnant women serve as additional barriers for drug manufacturers.

When drugs are approved, studies about whether they might cause birth defects are typically done only in animals, and they often don’t translate well to humans. So when a new medication comes on the market, nothing is known about how it affects people during pregnancy. Even if animal studies or the medication’s mode of action raised concerns, the drug can still be approved, though companies may be required to conduct studies observing its effects when taken during pregnancy.

Cause and effect

Of 290 drugs approved by the FDA between 2010 and 2019, 90% contain no human data on the risks or benefits for pregnant patients. About 80% of some 1,800 medications in a national database called TERIS, which summarizes evidence on medications’ risks during pregnancy, lack or have limited evidence about the risks for birth defects. Researchers have estimated that it takes 27 years to pin down whether a medication is safe to use in pregnancy.

As a result, many pregnant women stop treating their chronic diseases. In a U.S. study published in 2023, over one-third of women stopped taking a medication during pregnancy, and 36.5% of those did so without advice from a health care provider. More than half cited concerns about birth or developmental defects as the reason.

Yet uncontrolled chronic disease comes with its own toll on both the mother’s and the baby’s health. For example, some medications used to treat seizures are known to cause birth defects, but stopping them may increase seizures, which themselves raise the risk of fetal death.

Women with severe or recurrent depression who abruptly stop their antidepressants risk their depression returning, which is in turn associated with increased risk of substance use, inadequate prenatal care and other negative effects on fetal development. Stopping the use of medications
for treating high blood pressure also causes adverse effects – specifically, a greater risk of pregnancy-related high blood pressure that can cause organ damage, called preeclampsia; a condition called placental abruption, when the placenta detaches from the wall of the uterus too early; preterm birth; and fetal growth restriction. An online resource called Mother to Baby, created by a network of experts on birth defects, provides an excellent summary of the available data on medication safety during pregnancy.

The FDA in some cases requires drug companies to establish registries to track the outcomes of pregnancies exposed to certain medications. These registries can be useful, but they have shortcomings. For example, recruiting pregnant patients into them takes time and considerable effort, resulting in small sample sizes that may not capture rare birth defects. Also, registries typically follow a single medication and rarely include comparisons to alternative treatment approaches – or to no treatment.

What’s more, following the 2022 Dobbs v. Jackson Supreme Court decision overturning the constitutional right to abortion, women might be reluctant to add their names to a pregnancy registry or to provide data on prenatal detection of birth defects due to concerns about privacy and legal risks.

Decades of underfunding

In 2019, a task force established by the 21st Century Cures Act identified a major gap in knowledge about drug safety and effectiveness in pregnant and lactating women and recommended a boost in funding to fill it.

However, little has changed. A 2025 review by the National Academies of Sciences, Engineering and Medicine pointed out that research funding for women’s health topics has remained flat over the past decade, while the overall budget of the National Institutes of Health has steadily increased. The review recommended doubling the NIH funding allocated for such research, but this seems unlikely in light of the recent proposals to cut the overall NIH budget by 40%.

The National Institute of Child Health and Human Development funds the bulk of research on the safety of medications during pregnancy across federal agencies, although the institute has an appreciably smaller budget than most of its sister institutes such as the National Cancer Institute. Grants awarded are typically broad and take four to five years to complete, but they allow the more comprehensive assessments that are needed to support informed decisions considering outcomes for mother and child. For example, NIH-funded researchers have established a clear link between autism and prenatal use of valproate, a potent teratogen used to treat epilepsy and several mental health disorders.

The Centers for Disease Control and Prevention as well as the FDA have also funded specific pregnancy-related research. For example, following the COVID-19 epidemic, the CDC renewed its funding for studies that help expedite pregnancy safety studies for treatments that might be used for newly emerging infections. In response to emerging concerns about a substance called gadolinium, which is often used during MRI procedures, the FDA funded our own work on a study of almost 6,000 pregnant women, which found no elevated risk.

For healthy pregnancies, more research is critical

These efforts have laid a crucial foundation for evaluating medication safety and effectiveness during pregnancy. But keeping pace with the release of new medications and new ways they are used, as well as addressing the backlog of missing evidence for medications that were approved in the past millennium, remain a challenge.

Recent terminations of NIH-funded studies have focused on topics presumably relating to diversity, equity and inclusion. But research on safe and healthy pregnancies and on maternal health – for example, on the safety of COVID-19 vaccines during breastfeeding – has been affected as well.

The NIH has scaled back new grant awards by nearly US$5 billion since the beginning of 2025, and the odds for receiving NIH funding have plummeted. Proposed sweeping budget cuts for the CDC and FDA leave their role in supporting research on healthy pregnancies similarly uncertain.

In our view, removing or reducing ongoing investments in healthy pregnancies poses a danger to much-needed efforts to reduce excessive rates of stillbirths as well as infant and maternal deaths.

Almut Winterstein, Distinguished Professor of Pharmaceutical Outcomes & Policy, University of Florida and Sonja Rasmussen, Professor of Genetic Medicine, Johns Hopkins University

This article is republished from The Conversation under a Creative Commons license. Read the original article.