# I Know Why Lying about AI Water Use is So Easy Source: https://www.youtube.com/watch?v=H_c6MWk7PQc Recap page: https://rapidrecap.app/video/H_c6MWk7PQc Generated: 2025-12-08T23:27:57.176+00:00 --- ## Quick Overview The reason lying about AI water use is easy is that the actual life cycle water use for training models is complex and often hidden, leading to misleading comparisons where AI water use seems small compared to massive agricultural water use, while ignoring the pollution impact of thermal power plants used for AI electricity generation and the fact that AI companies are not transparent about their water sources. **Key Points:** - Sam Altman's claim that a ChatGPT query uses 1/15th of a teaspoon of water is highly misleading because it ignores the vast water used in the training phase of the AI models. - US Corn production alone requires approximately 20 trillion gallons of water annually, which is about 80 times more water than the total estimated global AI server water use (200 billion gallons) when comparing usage per unit of output. - Thermal electric power plants, which generate electricity for AI data centers, account for 45% of all fresh water withdrawals in the US, emphasizing that cooling water use is a major factor often omitted in AI water discussions. - The water used for cooling AI data centers is typically municipal water, which is often treated to drinking water quality, unlike industrial water used for power generation, which is often self-supplied and returned to local water bodies with thermal pollution. - The total environmental impact of AI is complex, involving not just direct water use but also energy consumption (leading to large carbon budgets) and the environmental impact of cooling water discharge that can harm local ecosystems. - The water used to train large models like GPT-5 is not directly counted in query estimates; instead, the training process—which is energy-intensive—is often separated from the query process, obscuring the true resource cost. ![Screenshot at 1:07: The speaker argues that the public narrative surrounding AI water use is misleading because it focuses only on query use while ignoring the massive energy and water footprint of model training and the source of the electricity used.](https://ss.rapidrecap.app/screens/H_c6MWk7PQc/00-01-07.png) **Context:** The video analyzes a claim made by Sam Altman regarding the minimal water usage of an individual ChatGPT query, contrasting this small figure with the much larger, often hidden, water consumption associated with the entire lifecycle of training large language models (LLMs). The speaker critiques the transparency of AI companies like OpenAI and discusses the broader resource implications of massive AI infrastructure, particularly in terms of electricity generation and water withdrawal from local sources. ## Detailed Analysis The speaker refutes the claim by Sam Altman that an average ChatGPT query uses only about 1/15th of a teaspoon of water (0.000085 gallons), arguing that this figure is misleading because it ignores the vast water consumption required for training the AI models themselves. The speaker points to research estimating that training models like GPT-5 requires hundreds of trillions of gallons of water, which is far more than what is used for queries. He contrasts this with US corn production, which uses 20 trillion gallons annually for irrigation, noting that 40% of this is used on corn that is fed to livestock, not humans. Furthermore, the speaker highlights that thermoelectric power generation, which supplies the electricity for AI data centers, accounts for 45% of all freshwater withdrawals in the US, emphasizing the massive environmental cost hidden behind the 'clean' water use metric. The water used for cooling AI data centers (often municipal water) is treated differently than industrial water used in power plants, which is often self-supplied and returned with thermal pollution. The speaker concludes that the focus on query water use is a political distraction, as the true environmental cost lies in the massive energy demand for training and the use of water near ecologically sensitive areas, which is rarely factored into the public discourse. ### AI Water Use Claims - Sam Altman claimed a query uses 1/15th of a teaspoon of water - This ignores the massive water used in training - Projections show AI use reaching 1,000 billion liters by 2028. ### Comparison to Agriculture - US corn production requires 20 trillion gallons of water annually - 40% of this corn is fed to livestock, not humans - AI water use is small compared to irrigation water use. ### Thermal Power's Role - Thermoelectric power generation accounts for 45% of US freshwater withdrawals - AI electricity demand contributes significantly to this, requiring water for cooling. ### The Real Problem - Water used for cooling AI systems is often municipal and treated to drinking quality, unlike industrial water - This comparison is misleading because the source and treatment differ significantly. ### Complexity and Incentives - The inherent complexity of resource analysis makes it easy to mislead the public - AI companies benefit from selective reporting (e.g., only counting query use, not training use) to downplay environmental impact. ![Screenshot at 00:07: Article headline stating Sam Altman claims an average ChatGPT query uses 'roughly one fifteenth of a teaspoon' of water.](https://ss.rapidrecap.app/screens/H_c6MWk7PQc/00-00-07.png) ![Screenshot at 00:22: Speaker emphatically stating the projected annual water use for AI data centers by 2028 will reach a trillion liters.](https://ss.rapidrecap.app/screens/H_c6MWk7PQc/00-00-22.png) ![Screenshot at 01:41: Ground News interface demonstrating how various outlets cover the same story with different biases and factual ratings.](https://ss.rapidrecap.app/screens/H_c6MWk7PQc/00-01-41.png) ![Screenshot at 04:06: Diagram illustrating a thermal power plant cooling system, highlighting the use of water for cooling and evaporation.](https://ss.rapidrecap.app/screens/H_c6MWk7PQc/00-04-06.png) ![Screenshot at 11:17: Pie chart from the USGS showing that thermoelectric power accounts for 45% of US water withdrawals, followed by irrigation at 32%.](https://ss.rapidrecap.app/screens/H_c6MWk7PQc/00-11-17.png)