Harvard expands AI food waste tracking across dining halls

Harvard University Dining Services has fully implemented Winnow, an AI-powered food waste monitoring system, across all residential dining halls and Annenberg. The rollout began last spring after a pilot program last fall. As of September, the technology had helped reduce food waste by 23,500 pounds, according to a report shared by Winnow.

The system uses a scale and overhead camera to weigh and photograph discarded food, then identifies items using data from dining services’ weekly menus. Daily and weekly reports rank the most-wasted foods by volume and cost, helping chefs adjust preparation. After discovering that they were discarding an average of 11.5 pounds a day of eggs, chefs began cooking smaller batches. Daily egg waste subsequently fell to 3.1 pounds, according to dining services’ yearly Learning Report.

Winnow representatives also meet regularly with lead chefs and dining services’ culinary operations director to identify broader patterns. Using that guidance, Harvard shortened its summer menu cycle, serving popular dishes more frequently and reducing low-demand offerings. Average food waste per meal subsequently fell 31 percent, from 2.12 ounces to 1.46 ounces per swipe, according to Winnow. The initiative complements Harvard’s reusable takeout container program, while some students stressed that diners must also avoid taking more food than they can eat.