Artificial Intelligence (AI) has become a vital part of our daily lives, influencing both industries as well as our own personal lives. In the workforce, AI assists with various tasks like writing, solving mathematical equations, and generating artwork. According to research by McKinsey, 55% of organisations have already started implementing AI in one way or another. This widespread adoption highlights the growing reliance on AI across various sectors, from healthcare and finance to manufacturing and retail.
Outside of work, AI enhances our everyday lives in numerous ways, making it an essential part of modern society. AI-powered virtual assistants like Siri, Alexa, and Google Assistant help people manage their schedules, set reminders, and quickly search for information. Additionally, AI applications can also help with more casual things, such as looking up recipes for banana pudding or turning our selfies into looking like sci-fi characters. AI is now something millions of people around the world use for all sorts of things, from making everyday tasks at work easier to casual and silly reasons as well.
However, the advent of new technology often brings a range of issues, and AI is no exception. Since its introduction, AI has raised numerous concerns, ranging from privacy issues—how much of our data is being analysed by AI bots—to more dramatic worries about AI becoming sentient and taking over the world, like Skynet from the Terminator movies. Yet, there is one significant issue that the public is not widely aware of: the enormous amount of freshwater required to run these AI systems. This critical concern underscores the need for greater awareness and sustainable practices in the development and deployment of AI technologies.
🥵 Why is AI So Thirsty?
Most people who use AI don't realise that running models like ChatGPT or Midjourney requires a lot of water. This is because AI needs a ton of energy, especially when training and using big models like GPT-3 and GPT-4. AI models learn from huge datasets, which involves processing a massive amount of information. This process uses parameters, which are variables the AI adjusts as it learns. These models have billions of parameters and are trained on groups of servers with lots of GPUs. These servers use as much energy as entire households, which creates a lot of heat. To keep them from overheating, these servers need to be cooled down, and that requires a significant amount of water.
The water consumption associated with running AI can be categorised into two ways: onsite server cooling and offsite electricity generation. Onsite cooling systems, such as cooling towers and liquid cooling mechanisms, require large volumes of water to remove heat from the servers. Additionally, the electricity needed to power these servers is often generated using methods that require water, such as thermal and nuclear power plants. This dual water usage highlights AI's substantial environmental impact, encompassing energy consumption, carbon emissions, and water usage.
When we look at how much water AI uses, it's important to understand the scale involved in training a single AI model, like GPT-3. Training one of these models can use up millions of liters of fresh water. Every time the model is used after that, it can consume about 500 mililitres of water. This large water footprint becomes even more concerning when you consider the rapid growth in demand for AI computing, which is expected to lead to a significant increase in water usage for AI in the coming years.
AI operations consume a lot of water, but data centers in general are also major water users. For example, Google's data centers are known for their high water consumption. In 2021, Google's data centers in the U.S. alone used an estimated 12.7 billion litres of freshwater just for on-site cooling. To give you an idea of the scale, about 90% of this water was potable, meaning it was clean enough for people to drink. This massive water usage is necessary to keep the servers cool and functioning properly, preventing them from overheating due to the intense energy they use. As more companies rely on data centers for AI and other computing needs, this water consumption is likely to grow, raising important questions about sustainability and resource management.
🤖 The Robots Aren’t the Only Culprit
It's equally crucial to recognise that while AI operations and data centers indeed consume significant amounts of water, they are not the sole industries contributing to freshwater depletion.
Agriculture 🥬
Agriculture stands as the largest consumer of water globally, utilising approximately 70% of freshwater resources. The extensive water use in agriculture is mainly for irrigation, fertilisation, pesticide application, crop cooling, and frost control. This heavy reliance on water is compounded by the pollution caused by fertilisers and insecticides, which further strain freshwater resources.
Among the thirstiest crops are wheat, corn, rice, cotton, and sugarcane. Nuts are also a significant concern, especially since 74% of irrigated nuts are grown in regions already facing water stress, such as India, China, Pakistan, the Mediterranean area, and the US. These crops require vast amounts of water, putting additional pressure on already scarce resources.
Fashion 👗
The fashion industry is also a major consumer of water, using around 93 billion cubic metres annually—enough to meet the needs of 5 million people. This number is expected to double by 2030, highlighting the growing water demand in this sector.
Water is integral throughout the garment production process. It starts with growing the plants and raising the animals needed to make fabrics. For instance, the cotton required to produce a single garment can need thousands of liters of water. Additionally, the extensive use of pesticides and insecticides in cotton farming leads to these chemicals washing into waterways, polluting ecosystems.
The manufacturing stage further contributes to the fashion industry’s water footprint. Fabric dyeing and treatment processes generate large amounts of wastewater. Often, this wastewater is untreated and pumped back into water systems, contaminating them with toxins and heavy metals. This not only affects water quality but also harms aquatic life and human health.
So, while AI and data centers do use a lot of water, they’re not the only ones to blame. Agriculture and fashion, among other industries, are also major players in depleting our freshwater resources. With the global population growing and consumer habits changing, the demand for these water-intensive products is only going up. This means even more strain on our precious water resources in the future.
👾 Tech to the Rescue
Despite these concerns, there are promising initiatives from tech giants aimed at reducing their water usage. Major tech companies, particularly those leading innovations in AI advancements, have recognised the urgent need to address the water crisis and have announced initiatives to mitigate their impact. These companies are increasingly aware of their environmental responsibilities and are taking steps to promote sustainability.
Microsoft
Microsoft has pledged to become "water positive" by 2023. This ambitious goal involves reducing their water consumption and replenishing more water than they use, positively impacting freshwater resources. Their strategy includes investing in wetland restoration and removing impervious surfaces like asphalt to help replenish water in highly stressed basins in locations that they operate.
They're also leveraging technology to identify water-stressed areas and optimise replenishment efforts. Through their AI for Earth program, Microsoft supports projects like the Leadership Counsel for Justice and Accountability in California's Central Valley. This nonprofit uses AI to predict and prevent domestic well failures, helping local agencies improve water management for the 1.5 million people relying on these wells.
Additionally, Microsoft is implementing water-saving technologies at various campuses worldwide, such as rainwater collection systems and wastewater recycling. They are also partnering with NGOs to provide over 1.5 million people with access to clean drinking and sanitation water in seven countries. This initiative reflects a comprehensive approach to addressing both water availability and accessibility.
Google has been making significant strides to address the water crisis. In 2021, they made a bold declaration: by 2030, they're aiming to replenish 120% of the freshwater they consume, spanning their offices and data centers. It's a substantial commitment that underscores Google's dedication to environmental stewardship. By striving to return more water than they use, Google is taking proactive steps to safeguard our water resources for future generations.
One of their initiatives involves revitalising floodplains in California’s Central Valley, an area increasingly impacted by climate change-induced droughts and floods. Teaming up with River Partners, Google has taken steps to combat this challenge. They've secured riverside farmland, which suffered severe flooding in 2023, devastating the crops. River Partners is set to restore the floodplain by planting thousands of native trees and plants and reconnecting it with the Tuolumne River. This restoration effort aims to support endangered species like Chinook salmon and steelhead trout, reduce water demand, recharge local aquifers, and enhance flood safety for the community.
🌏 Time to Buck Up
AI’s heavy use of freshwater is a serious issue that needs immediate attention, beginning with global authorities and industry leaders. With climate change worsening water scarcity, experts predict that by 2030, global water demand will exceed supply by 40%.
As AI becomes more embedded in our lives, its water consumption will grow. Raising awareness about AI’s water use can help push tech companies to be more sustainable. Google, for instance, is investing in building data centers in Malaysia. While this will boost the economy and create jobs, the government must monitor their water use to ensure it doesn't further harm the environment.
The United Nation’s Sustainable Development Goal (SDG) 6 aims to "ensure availability and sustainable management of water and sanitation for all," but we are far from achieving this. Currently, 1 in 4 people lack access to safe drinking water. If AI data centers continue to operate as they do now, this trend will worsen. Therefore, it’s crucial to ramp up efforts for sustainable water management.
Comments