Featured Analysis: Sam Altman claims an average ChatGPT query uses ‘roughly one fifteenth of a teaspoon’ of water
This is a summary and commentary on the article ‘Sam Altman claims an average ChatGPT query uses ‘roughly one fifteenth of a teaspoon’ of water’.
Summary
OpenAI CEO Sam Altman stated in a recent blog post that an average ChatGPT query consumes approximately 0.000085 gallons of water (roughly one-fifteenth of a teaspoon) and 0.34 watt-hours of energy. He compared the energy usage to that of a household oven or lightbulb, and predicted that the cost of AI intelligence will eventually approach the cost of electricity. This claim comes amidst growing concern over the environmental impact of AI, with some research suggesting AI’s energy consumption could surpass that of Bitcoin mining. Previous studies have highlighted the significant water usage of AI, varying based on datacenter location, with one example showing a 100-word AI-generated email consuming over a bottle of water. OpenAI did not immediately provide details on how Altman arrived at his figures.
Commentary
Altman’s claims regarding ChatGPT’s resource consumption are noteworthy given the increasing scrutiny surrounding AI’s environmental footprint. While the presented figures seem low, the lack of transparency in their derivation raises concerns about their validity and methodology. The significant variation in previous research on water usage underscores the complexity of assessing AI’s environmental impact. This complexity stems from multiple factors, including datacenter location, energy sources, and the specific AI model used. Altman’s prediction about the cost of intelligence converging with the cost of electricity, while aspirational, hinges on significant technological advancements in energy efficiency. The broader implication is that the environmental sustainability of AI needs rigorous and transparent investigation, moving beyond isolated claims and towards a comprehensive understanding of its resource demands across the lifecycle. This requires collaboration between AI companies, researchers, and regulators to establish standardized measurement methods and promote the development of more energy-efficient AI systems.
本文内容主要参考以下来源整理而成：