Food waste: AI data helps Harvard dining cut waste per meal 31% with a shorter menu cycle
Harvard University Dining Services has fully rolled out Winnow, an AI system that weighs and photographs food scraps and identifies items from the weekly menus, across all residential dining halls and Annenberg after a pilot last fall. Per Winnow's report, the system had helped cut 23,500 pounds of food waste as of September. Using the data, chefs moved eggs to smaller batches (daily egg waste fell from 11.5 to 3.1 pounds), and HUDS shortened its menu from a four-week to a three-week cycle, after which average waste per meal fell 31%, from 2.12 to 1.46 ounces per swipe, according to Winnow's report.
Why it matters for directors: Waste data saves money only when it changes production and the menu itself. Track waste per meal served, then cut or rotate the dishes students leave behind.
Original source: The Harvard Crimson ↗