Environmental Costs of AI Advancements: In the quest to harness the power of artificial intelligence, tech giants like Microsoft, OpenAI, and Google have embarked on a journey that carries hefty environmental costs, including increased water consumption. While the cost of building AI products like ChatGPT can be challenging to measure, the environmental impact is becoming increasingly apparent.
The Hidden Water Footprint of AI
One critical aspect of developing advanced AI models is the need for vast amounts of computational power. To facilitate this, supercomputers are employed to train AI systems like ChatGPT. However, these supercomputers generate substantial heat, necessitating cooling systems that rely on water. This crucial cooling process often takes place in large data centers, where immense computing power is concentrated.
In their race to capitalize on the generative AI trend, tech leaders are generally tight-lipped about the specifics of their environmental impact. Still, recent revelations shed light on the significance of water consumption. For instance, Microsoft disclosed in its latest environmental report that its global water consumption surged by 34% from 2021 to 2022, primarily due to its investments in AI, including its partnership with OpenAI.
A Thirsty AI
A forthcoming research paper by Shaolei Ren and his team at the University of California, Riverside, provides insight into the water usage of ChatGPT. According to their estimates, asking a series of 5 to 50 prompts or questions to ChatGPT consumes approximately 500 milliliters of water, equivalent to the contents of a 16-ounce water bottle. These estimates encompass both direct and indirect water usage, such as water required to cool the power plants that supply data centers with electricity.
Google also reported a 20% increase in water usage, which Ren attributes to its AI-related projects. This surge wasn’t uniform, with notable increases in regions like Oregon and Iowa, where Google’s data centers drew substantial amounts of water.
The Birthplace of AI Advancement
West Des Moines, Iowa, may not be the first place that comes to mind when you think of AI innovation. Still, it played a crucial role in OpenAI’s most advanced large language model, GPT-4. Microsoft invested heavily in building powerful supercomputers in this city to support AI model training. This development was shrouded in secrecy until Microsoft’s president, Brad Smith, revealed the location, stating that “it was made in Iowa.”
Iowa’s climate offers an advantage in terms of water usage. The cool weather for most of the year allows Microsoft to use outside air for cooling, reducing the reliance on water-based cooling systems. However, when temperatures exceed 85 degrees Fahrenheit, water becomes a necessity.
Addressing Environmental Concerns
Recognizing the environmental impact of their AI initiatives, both Microsoft and OpenAI have expressed their commitment to sustainability. They are investing in research to measure AI’s energy and carbon footprint while seeking ways to enhance the efficiency of large-scale AI systems.
In conclusion, as the demand for generative AI continues to grow, it is crucial to consider the environmental costs associated with these advancements. Awareness of the water usage and other resource consumption underlying AI models like ChatGPT is essential to ensure responsible development and conservation of resources in the pursuit of technological innovation.
Also read:
Chrome’s Controversial “Privacy Sandbox” Ad Platform Quietly Rolls Out
