AI and Food

Rewind a few years and in the middle of COVID lockdowns an interaction designer and a computational materials scientist—that’s us—started talking not about the technical aspects of food and AI, like how AI might reduce water and pesticide usage in agriculture, optimize supply chains, or conjure new exciting flavours, but about deeper, more fundamental ones. Who gets to build and operate these AI models? What biases are inherent in them? How transparent are they? Who are they designed to benefit? How should they be legislated? In short, how will AI be implemented in food systems to serve the world? No pun intended.

What was missing in these dazzling showcases of AI’s potential was a critical conversation about food—the very foundation of human and planetary health. This absence becomes even more glaring when you consider the fragile and failing state of today’s food systems.

People around the world are more disconnected from their food than ever before. They often don’t know where it comes from, how it’s grown, or who grew it. Modern food systems, driven by profit and convenience, rely on practices that are extractive and damaging to people and planet. Indeed many now describe global food systems as being broken. Climate change, food insecurity, biodiversity loss and ecosystem erosion are already upon us and steadily worsening. Food systems are now so opaque and complex that making informed decisions about what to eat can feel, or even legitimately be, impossible. With these challenges intensifying, we must rethink how food is grown, distributed, and consumed, and create food systems that are equitable, sustainable, resilient and transparent. 

Technology, we are often told, offers the solution. Innovate, innovate, innovate is the clarion call!  To this end, the rapid advance of AI presents incredible opportunities, alongside significant risks. From precision agriculture and personalised nutrition, to helping us understand the real environmental and social costs of our food choices, we know that AI will change global food systems. But there’s also the risk that AI will be used to prioritise convenience and profit, repeating the same mistakes that have led to monoculture farming and ecosystem devastation. How do we steer new systems away from entrenching existing issues and towards alleviating them?

We had attended the conference with the singular mission of initiating conversations around this. And when we spoke with attendees, responses were often bemused: “AI and food? Fascinating, I haven’t heard anyone else here discussing that.”

As the conference wore on, a pattern emerged. The AI inventions showcased — awe-inspiring as they were — seemed to be made by and for a very specific demographic. Mostly older white male humans from affluent countries. All AI systems, to some extent, contain biases. (The old adage of “all models are wrong, but some are useful” springs to mind.) Biases creep into AI via datasets, the things they are trained on, or unconsciously via the humans who develop them. Either way, they emerge in the model’s predictions and outcomes, which might otherwise appear impartial and accurate. We need AI systems that are transparent, that openly acknowledge and mitigate their biases, and that involves many different people involved in their creation. Standing in the final hours of the conference, watching a team of male engineers dress a buxom female humanoid robot — that has received global coverage but also since been found to have rather limited truly AI capabilities, despite the hype — into a short skirt felt like a dystopian beauty pageant, a jarring reminder of how technology can inherit humanity’s oldest clichés. It became clear to us: many of the ethical implications and intersectional issues pertinent to the use of AI are being commonly overlooked by key players.

What was missing in some parts of that conference centre, and food and AI conversations more broadly, is a more inclusive vision. One that doesn’t just serve the needs of a few, but of many, especially those all too often excluded from the conversation: women, people of colour, and the farmers of the Global South who produce a lot of the world’s food (who are often women of colour). One in which not just human needs are considered but those of all the members of the ecosystems that make our food, the flora and fauna and the vast networks of microbes that nourish them. Our mission is to shift from “human-centred design” to “earth-centered design.” We want to create a food system that values the health of ecosystems just as much as it values human well-being. To do that, we must place the planet’s needs, in a holistic way, at the centre of every decision we make, ensuring that the food we consume nourishes both people and the Earth. 

—

We’ve been exploring the possibilities and challenges AI presents for food systems. Our original thought experiment, formed during the middle of COVID, focused on a simple question: if you could create an AI that could speak on behalf of all of planet Earth, what would you ask it, and what might it want to ask of you? We imagined an AI powerful enough to collect data from every bug and bacteria, every ocean and each grain of salt and sand within it, and every apple tree and apple, and factor their needs into its answers. Standing in a supermarket of the future you might ask it, “Should I buy this apple, or this pear?” 

That led us down a rabbithole of food and AI, and we are now bringing the fruits of some of those discussions to a global, communal stage in the form of the Augmented Food Studio. The potential of AI in food is too significant to be confined to labs or corporate boardrooms. It’s a conversation that needs to involve everyone—farmers, scientists, designers, and policy-makers alike. We’re creating a platform to spark an open and transparent global dialogue around AI and food. This platform will not only serve as a space for learning and knowledge-sharing but also as a catalyst for action. We will explore critical questions such as how AI can be designed to prioritise sustainability for the many over profit for the few; how it can create food systems that are resilient in the face of climate change; how AI-driven food innovations can be accessible to all and don’t deepen existing inequalities; how AI systems be designed to prioritise long-term planetary health; and what ethical considerations must guide the integration of AI into global food systems.

As a part of this mission, we’re thrilled to announce our collaboration with MOLD magazine, via a series of articles exploring the role of AI in food. These pieces will examine how AI might reduce food waste and revolutionize personalised nutrition, what changes consumers are likely to see due to AI in food, and how AI datasets can mirror human biases, risking the reinforcement of systemic inequalities in everything from ingredient choices to marketing strategies. We’ll also look at AI’s role in sustainability, analyzing its potential to both aid and hinder sustainable food systems, and discuss how AI can be developed to ensure equity, access, and circularity for all communities. Further, we’ll envision a future where AI not only serves human needs but also supports biodiversity and ecological resilience. 

In addition, we’re hosting a United Nations Webinar titled “Shaping Tomorrow’s Plates” on November 12. This event will bring together leading experts in AI, food innovation, and sustainability to discuss the future of food and how AI can help address global challenges. The webinar will explore how AI can revolutionise the way we grow, prepare, and consume food while addressing the ethical and environmental questions surrounding this technology. And everyone is invited. 

Whether you’re an artist, designer, teacher, cook, farmer, scientist, or foodie, we want you to join us on this journey as we explore the future of AI in food. As we stand on the brink of an AI-driven food revolution, the question isn’t just what AI can do for us, but what it can do for the planet. Together, we can ensure that tomorrow’s plates nourish all life—not just the privileged few.