Sustainability is an increasingly pressing topic worldwide, and AI is a conversation starter all too often. But what could these two concepts have in common? Companies around the world have begun to explore the growing connection between the two, and a recent study titled “Making AI Less Thirsty” delves into how much water and power are used when training AI models like ChatGPT and Bard.
The research team for this study includes academics from the University of Texas at Arlington and the University of Colorado Riverside. They have determined that water and power are required to keep the data centers running, with Microsoft using up to 185,000 gallons of water to train GPT-3. This would be enough to cool a nuclear reactor and adds up to approximately three times the amount of water if done in their Southeast Asia-based data centers.
The study aims to compare the required resources for AI training, for which water is an essential component, as it is needed to cool the data centers. During this process, much water is lost and not returned to the environment. This demonstrates the sheer demand for water while training powerful AI models such as ChatGPT and shows why sustainability remains a pressing concern. As mentioned in the paper, a mere 500 ml of water is consumed to respond to 20-50 questions asked of the AI, but when the entire data center is taken into account, the total water usage is immense.
It is critical that companies utilising data centers be aware of the effects of their operational activities on the environment, and ensure that the data centers are as sustainable and efficient as possible, as one can only imagine the amount of water being used for every AI-powered task. AI might be the technology of the future, but it should not come at the cost of using up water resources. There are several methods to help reduce water usage, from the deployment of air-cooled and water-cooled systems to the use of recycled or reclaimed water.
Microsoft is an American multi-national tech company and a leader in the development of cloud computing, artificial intelligence, and data science. Founded in 1975, it is one of the world’s most valuable companies and has been credited with revolutionizing the modern age. Microsoft’s co-founder and current CEO, Satya Nadella, is responsible for driving the company’s expansion and focus on artificial intelligence, cloud computing, and machine learning.
OpenAI is a research laboratory based in San Francisco, California and was established in December 2015 by a group of noted researchers, entrepreneurs, and engineers. Founded by tech giants such as Microsoft, Amazon Web Services, and Y Combinator, OpenAI was established with the goal of advancing artificial intelligence in the most responsible and powerful way possible. The organization has become well-regarded in the artificial intelligence space and its partnership with Microsoft has helped to drive the development of many AI models, including GPT-3.
