AI has a hidden water cost − here's how to calculate yours - The Conversation

An academic librarian explores the water consumption behind AI systems, revealing that ChatGPT uses up to 500 milliliters of water per conversation. The article examines how AI's environmental impact varies dramatically based on location, timing, and technology choices. Understanding these factors helps users make informed decisions about balancing AI innovation with sustainability concerns.

• Two water streams: AI consumes water through data center cooling systems and electricity generation at power plants, with usage varying significantly by climate and infrastructure
• Location matters: Data centers in cool climates like Ireland use minimal water, while those in hot, dry regions like Arizona consume much more through evaporative cooling
• Simple calculation method: Users can estimate AI water footprint using energy per prompt multiplied by water factor, revealing dramatic differences between AI models
• Context and solutions: Daily AI water use remains small compared to lawn watering, while emerging technologies like immersion cooling and closed-loop systems offer promising alternatives