Harvard's dining halls are grappling with a massive food waste crisis – but could artificial intelligence be the game-changer we've all been waiting for?
Imagine tossing out perfectly good meals every day, not just in your home kitchen, but on a university scale. That's the reality at Harvard, where administrators are turning to cutting-edge tech to tackle this pressing environmental issue. Harvard University Dining Services (HUDS) is currently testing an innovative AI-driven tool called Winnow in Adams House, Currier House, and Annenberg Hall. This system aims to slash food waste by analyzing what's being discarded before it even reaches consumers – a concept known as pre-consumer waste, which refers to scraps from preparation and serving that haven't been eaten yet.
But here's where it gets fascinating: Winnow combines a high-tech floor scale with a smart camera to weigh compost bin contents and capture images of non-donatable food scraps. The AI then meticulously identifies the exact menu item involved, providing real-time insights into what's going uneaten. For beginners, think of it like a super-smart detective that scans leftovers and tells you, 'Hey, that's extra rice from lunch!' This isn't just about tech for tech's sake; it's part of HUDS's wider commitment to sustainability, encompassing health, cultural, religious, and environmental considerations. As a result, they're rolling out other eco-friendly changes for the Fall 2025 semester, such as swapping disposable takeout containers for reusable ones – a small step that could make a big difference in reducing plastic pollution.
Smith Haneef, HUDS's managing director, explained the motivation behind this initiative: 'In terms of environmental awareness, food waste stood out as an area where we wanted to enhance our systems and processes for better efficiency and effectiveness.' HUDS sought a straightforward, precise way to address waste from their all-you-can-eat buffet model. After consulting with four or five companies offering similar technologies, directors Martin Breslin (culinary operations) and Crista Martin (strategic initiatives and communications) chose Winnow based on key factors.
'As we spoke with these firms, we selected Winnow for a few reasons, primarily because it needs to be user-friendly for our team and deliver highly reliable, precise data,' Breslin noted. And the results are already promising. By the third week of the pilot, HUDS collected initial data that helped them decide what foods to prepare and in what quantities to cut down on waste.
'Every day, when I check a Winnow report, I can instantly see what's being returned, allowing our kitchen staff to tweak production on the fly and cut back on excess,' Breslin added. Previously, they depended on manual, tedious handwritten logs, but Winnow automates this, making data collection simpler and far more accurate.
Of course, no system is flawless. And this is the part most people miss – the AI isn't perfect. 'Does Winnow's visual AI accurately recognize the ingredient or dish, or could it be mistaken, like confusing macaroni and cheese for chicken noodle soup?' Haneef questioned. There have been some inaccuracies, but the team counters this with human oversight at the end of each shift, creating a feedback loop to refine the AI and camera.
While Winnow's forecasting capabilities are still developing, the pilot has already revealed overproduced items across the three locations. 'Rice tops the list,' Breslin said. 'It showed up consistently, and fixing it was straightforward – almost instant.' Beyond predictive analytics that quantify what HUDS suspected intuitively, they've condensed menu rotations from four weeks to three, focusing on popular student dishes to boost efficiency.
'A longer four-week cycle introduces more variety, which can lead to more inventory and, consequently, more waste,' Breslin explained. 'Switching to three weeks keeps things streamlined and more effective.' Haneef is optimistic about expanding Winnow university-wide after a review on December 20, 2025, at semester's end.
'Our ambitious internal target is to hit a 15 percent reduction in food waste at the production stage,' she stated. 'If we achieve that, we'll convene a follow-up team to discuss scaling it further.'
But here's where it gets controversial: Is relying on AI to dictate menu portions and waste reduction the ultimate solution, or could it inadvertently compromise food quality or variety? Some might argue that over-emphasizing data could lead to bland, repetitive meals, prioritizing efficiency over culinary creativity. What do you think – should universities like Harvard lean harder on technology for sustainability, or explore more traditional methods like better portion control or community education? Do you agree that AI errors are just a minor hurdle, or do they raise bigger concerns about tech dependency? Share your opinions in the comments below – we'd love to hear your take!
— Staff writer Ava H. Rem can be reached at ava.rem@thecrimson.com. Follow her on X @avar3m.
