"training GPT-3 in Microsoft's state-of-the-art US data centers can directly consume 700,000 liters of clean freshwater", which they calculate could be used to produce 370 BMW cars or 320 Tesla electric vehicles.
Furthermore, ChatGPT 'drinks' the equivalent of a 500ml bottle of water for a simple conversation of 20-50 questions and answers. Which may not seem like much... until you consider that the chatbot has more than 100 million active users, each of whom engages in multiple conversations.
So, I thought it would be a worthy use of a little H2O to ask ChatGPT what it thought of the situation.
It was surprisingly frank:
My question: As a person interested in sustainability, I am alarmed to see that AI is very 'thirsty' for freshwater, based on this article https://www-forbes-com.cdn.ampproject.org/c/s/www.forbes.com/sites/federicoguerrini/2023/04/14/ais-unsustainable-water-use-how-tech-giants-contribute-to-global-water-shortages/amp/ . Is it true that AI consumes large amounts of water, and what is being done about it?
The answer from ChatGPT: Yes, it is true that AI systems consume significant amounts of water, especially in the training phase of machine learning models. This is because AI requires vast amounts of computational power, which generates heat that needs to be dissipated, and cooling systems that use water to maintain the necessary temperature. Additionally, data centers that host AI models require a lot of water for cooling and other operational needs.
The impact of AI on water resources is an important concern, and there are ongoing efforts to address this issue. One approach is to use more energy-efficient hardware, such as specialized processors that can perform AI computations with lower power consumption. Another approach is to develop more efficient cooling systems that use less water, and to use recycled water or other sustainable sources of water for cooling.
In addition to technical solutions, it's important to consider the environmental impact of AI in the design and deployment of AI systems. This means taking a holistic approach to sustainability and considering the entire life cycle of AI systems, including the raw materials used to manufacture hardware, the energy used to power AI models, and the disposal of electronic waste.
Overall, the AI industry and the broader technology sector are increasingly aware of the need to address the environmental impact of their activities, including water use, and are taking steps to reduce their impact on the planet.
The video just below from Amazon Web Services (AWS) tackles the issue and provides some ways that AWS is helping to battle this, wanting to be “water-positive” by 2030. When you watch this video, note that what they are really discussing, without saying it directly, is that they have a portfolio of programs and projects which initiatives that execute the strategy aimed at that goal, which is in turn, aligned with their sustainability-oriented mission and vision statement. This is very much in line with the model that (for example), folks like Gilbert Silvius, Alexandra Chapman, and others have been promoting for a long time.
Google recently shared a blog post revealing its water use – a staggering 4.3 billion gallons of water in 2021, which works out to an average daily water footprint approaching a half-million gallons per data center.
It gets complicated as well, because there is a trade-off (we’re used to those as a project manager, though) between reducing water and reducing carbon. The initiatives you see in the video, I’m sure you have already guessed, take energy and have their own impact. From the Google blog post:
When used responsibly, water cooling can play an important role in reducing emissions and mitigating climate change. Water-cooled data centers use about 10% less energy and thus emit roughly 10% less carbon emissions than many air-cooled data centers. In 2021, water cooling helped us reduce the energy-related carbon footprint of our data center portfolio by roughly 300,000 tons of CO2."
So it is a complicated, convoluted, higher-level, higher stakes balancing act that we must perform.
“Going Waterless” was one of the top eight trends identified by Data Center Frontier. So at this point it seems we’re at least conscious of the problem. Now – can we get the right amount of talented project, program, and portfolio leaders on the job? It’s an important enough problem to pour over…
