Unveiling the Thirsty Nature of AI: The Case of ChatGPT

JJohn September 9, 2023 3:41 PM

The creation of Artificial Intelligence systems like ChatGPT involves significant environmental costs, including increased water usage for cooling technology. This article explores the hidden environmental impact of AI and the industry's efforts to address these issues.

The water-intensive nature of AI development

Building an AI model isn't just about complex algorithms and vast amounts of data. It also needs a physical infrastructure to support the intense computing processes. The supercomputers used to develop models like ChatGPT, created by OpenAI with Microsoft's backing, generate a lot of heat. As a result, they require significant amounts of water to keep them cool. These cooling needs are so great that they've influenced the location of the development process, leading to Iowa's unsuspected status as the birthplace of OpenAI's advanced large language model, GPT-4.

Despite the clear environmental impact, many tech companies are reticent to disclose the full extent of their AI tools' resource usage. Few outside the industry were aware of Iowa's role in the development of the GPT-4 model before it was highlighted by a senior Microsoft executive. This lack of transparency is problematic for those trying to gauge the environmental effect of these tools. The University of California's Shaolei Ren, for instance, has noted the difficulty in calculating the water usage associated with AI models given the indirect water usage related to cooling power plants that the companies do not measure.

Rising water consumption in AI research

The environmental reports from tech giants Microsoft and Google present a clear picture of the rising water consumption related to AI work. Microsoft, for instance, noted a 34% increase in global water consumption from 2021 to 2022, a jump that researchers attribute largely to its AI research. Google reported a similar trend with a 20% growth in water use in the same period. This increase in water use isn't just an abstract concern but has real-world implications, as seen in places like Iowa where the data centers draw more potable water than anywhere else.

Efforts to address AI's environmental challenges

In the face of these environmental challenges, tech companies are making moves to become more sustainable. Microsoft, for instance, is investing in research to measure and reduce the energy and carbon footprint of its AI operations. The company is also working on ways to make large systems more efficient. OpenAI has mirrored these sentiments, acknowledging the energy and water-intensive nature of training large models and committing to improving efficiencies. These efforts are crucial for the industry to move towards a more sustainable future.

More articles

Also read

Here are some interesting articles on other sites from our network.