AI water-use critique

- Digital Camera World published a commentary calling on editing-software companies to disclose how much water their AI features consume. - The piece emphasized uncertainty about water usage tied to AI features and urged transparency for Earth Day. - The article signals growing reputational scrutiny of AI tools' environmental costs, beyond pure performance debates. (digitalcameraworld.com)

A photography site used Earth Day to press editing-software makers to say how much water their artificial-intelligence tools consume. (digitalcameraworld.com) The April 22 commentary in Digital Camera World said photo apps now ship with more artificial-intelligence features, but users still do not get clear water-use figures for those tools. (digitalcameraworld.com) In photo editing, many of those features run in remote data centers, not just on a laptop. Adobe markets Photoshop Generative Fill as a text-prompt tool that can add or remove content, and Lightroom’s Generative Remove went out to users in May 2024. (adobe.com) (news.adobe.com) Water enters the story because data centers use it for cooling, the same way a radiator carries heat away from an engine. Microsoft says water usage effectiveness tracks liters of water used for cooling and humidification against the electricity consumed by information-technology equipment. (microsoft.com) Microsoft’s datacenter site says its global water usage effectiveness was 0.27 liters per kilowatt-hour in fiscal year 2025, down from 0.30 in fiscal year 2024. The company says those figures cover datacenters it fully owns and controls and that were operational for 12 months in the reporting period. (microsoft.com) What users still cannot easily see is a per-feature number for one Photoshop fill, one Lightroom removal, or one Canva Magic Edit request. Canva’s help page explains how Magic Edit works, but it does not publish a water figure alongside the tool instructions. (canva.com) Researchers have tried to estimate the gap. A 2025 revision of a widely cited paper said training GPT-3 in Microsoft data centers could directly evaporate 700,000 liters of freshwater, and projected global artificial-intelligence demand could withdraw 4.2 billion to 6.6 billion cubic meters of water in 2027. (arxiv.org) A 2025 Nature Sustainability paper put the uncertainty in newer terms, estimating artificial-intelligence servers in the United States could add an annual water footprint of 731 million to 1.125 billion cubic meters between 2024 and 2030, depending on how fast deployments grow. The paper said location, grid mix, and efficiency measures all change the total. (nature.com) Adobe does publish broader sustainability targets, including net zero by 2050 and a 25% reduction in global water usage per full-time employee by 2025. Its Responsible AI page also says the company is guided by accountability, responsibility, and transparency when it evaluates new AI features before release. (adobe.com 1) (adobe.com 2) Microsoft has also tried to answer part of the criticism at the infrastructure level. In December 2024, it said a next-generation datacenter design launched in August 2024 uses zero water for cooling, though that does not translate into a public water label for each consumer AI feature. (microsoft.com) The Earth Day complaint was narrow: not a demand to remove AI tools, but a demand to label them. The missing number is still the same one the commentary started with — how much water one edit actually uses. (digitalcameraworld.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.