Utilizing AI to transform oneself into a 1980s movie icon is currently a popular trend. By entering a prompt and uploading a photo, users can swiftly obtain a vintage, film magazine-worthy portrait of themselves. This process, which evokes nostalgia and fun, appears effortless on the surface.
Yet, the seemingly simple act of generating retro images using AI conceals a significant environmental impact. Behind the facade of the “generate” button lie extensive data centers filled with servers that consume substantial amounts of electricity and water. Despite this hidden infrastructure, the environmental consequences of AI usage should not be disregarded.
AI technology is employed for various tasks such as writing, editing, generating content, and image manipulation. Each command given to AI, no matter how trivial it may seem individually, collectively contributes to substantial resource consumption.
The operation of generative AI relies heavily on significant computing power, requiring specialized chips that consume electricity and produce heat. Cooling systems utilized to manage this heat further consume substantial amounts of water. The water used for cooling also plays a role in generating electricity to power data centers, with resource demand escalating for more computing-intensive tasks.
The exact water consumption per prompt varies based on numerous factors such as model, data center specifications, cooling methods, electricity sources, and the nature of the request. Estimates suggest varying water usage levels, with reports indicating a wide range from different sources.
Despite the growing concern over global water scarcity, AI companies lack standardized transparency regarding their water footprint. Projections indicate a substantial annual global water footprint from AI technology, emphasizing the critical need for increased awareness and sustainable practices.
Balancing the benefits of AI with environmental considerations is crucial. Efforts should focus on enhancing disclosure practices by AI companies, promoting water-efficient data centers, utilizing recycled water, and implementing regulations on data center locations in water-stressed areas.
Users are encouraged to question the necessity of using AI for each task to minimize unnecessary resource consumption. Acknowledging and addressing the environmental impact of AI operations is essential for sustainable technological advancement.
Hurmetun Nesa Labiba and Fatema Tuz Zuhra are affiliated with the Bangladesh Institute of Governance and Management (BIGM).
Opinions expressed in this content belong to the authors.
For the latest opinions and analyses, follow The Daily Star Opinion on Facebook. To contribute articles or letters, refer to our submission guidelines.
