STAT+: The real story behind AI water usage
\n \t
\n
<\/BR>
\n
\n \t
\n
<\/BR><\/p>\n
You\u2019re reading the web edition of STAT\u2019s AI Prognosis newsletter, our subscriber-exclusive guide to artificial intelligence in health care and medicine.\u00a0Sign up to get it<\/a>\u00a0delivered in your inbox every Wednesday.<\/em>\u00a0<\/p>\n As you\u2019ll see, AI Prognosis looks a little different today because I\u2019ve been busy attending and traveling to\/from the National Association of Science Writers\u2019 conference in Chicago. Keep reading for some fun AI-related things I learned there.<\/p>\n As always: Email tips, questions, story ideas to aiprognosis@statnews.com<\/a>, or contact me on Signal at btrang.01 for sensitive story tips.<\/p>\n A conference session I attended this weekend addressed the climate and health effects of data centers. What caught my attention was an explanation of why Google says<\/a> that its AI uses only 0.26 mL of water (or five\u00a0drops) per prompt, whereas French AI company Mistral says<\/a> its AI model uses 45 mL.\u00a0<\/p>\n Shaolei Ren, an electrical and computer engineering professor at the University of California, Riverside, broke down electricity\/water usage at data centers into three parts:\u00a0<\/p>\n Continue to STAT+ to read the full story\u2026<\/a><\/p>\n<\/div>\nHow much water does AI really<\/em> use?<\/h2>\n