40,2601$% 0.13
46,7458€% 0.13
53,9601£% 0.23
4.316,24%0,46
3.337,10%0,40
10.198,76%-0,26
In a recent blog post published this week, OpenAI CEO Sam Altman shed light on the surprisingly minimal water footprint associated with individual ChatGPT queries. He explained that each query roughly equates to just 0.000085 gallons of water, which is approximately one fifteenth of a teaspoon. Altman’s detailed explanation was part of a broader initiative to address concerns about the environmental implications of artificial intelligence technologies.
Altman further elaborated on ChatGPT’s energy consumption, revealing that an average query consumes about 0.34 watt-hours of electricity. To put this into perspective, this amount of energy is comparable to what a typical stove uses in just over a second or the electricity consumed by a low-energy light bulb over a few minutes. He emphasized the long-term goal that the operational costs of AI, in terms of electricity, should eventually align with the costs of electricity itself, suggesting a future where AI efficiency becomes more economically sustainable.
Despite these intriguing insights, OpenAI has yet to provide comprehensive details about the methodologies used to calculate the water and energy metrics. Requests for further clarification or detailed data have not been answered promptly, leaving some questions about the precision and applicability of these figures.
The increasing energy demands of artificial intelligence systems continue to attract scrutiny from environmental advocates and researchers. Earlier this year, some experts warned that AI could surpass Bitcoin mining in energy consumption by the end of the year. Additionally, a Washington Post article from last year highlighted that creating a simple 100-word email might consume as much water as a standard water bottle. The study also pointed out that water consumption levels can vary significantly depending on the data center’s location, further complicating the environmental impact assessments.
Mastering Prompt Crafting: How to Get the Most Accurate Answers from ChatGPT