University of Leeds researchers have collaborated with scientists from CSIRO, Australia’s national research organization, to develop an artificial intelligence–based solution to reduce food waste and strengthen global food security.
Image Credit: Fevziie/Shutterstock.com
What is Fermentation?
Fermentation has long been used to preserve and transform food - from bread and cheese to wine and yogurt. In recent decades, this biological process has shown potential to convert waste into valuable products such as microbial protein, which can serve as an ingredient in food production or animal feed. However, implementing fermentation at scale for highly variable waste streams has proven complex.
Differences in composition, moisture, and nutrient content can change how microbes behave, making it difficult to establish consistent conditions that yield high-quality protein at a competitive cost. Consequently, even though upcycled protein has sustainability benefits, its economic viability relative to traditional protein sources remains a barrier to adoption by industry.
Recognizing these challenges, the University of Leeds and CSIRO have joined forces to create an AI-powered platform that can navigate the complexity of waste fermentation. The effort is funded by the Bezos Earth Fund’s AI for Climate and Nature Grand Challenge, a global initiative that supports AI solutions for climate resilience, biodiversity conservation, and food security.
Why is the AI System Being Developed?
At the heart of this project is an AI system that integrates data from diverse fermentation experiments and real-world use cases to recommend optimal process conditions. The tool is being designed to account for the high variability of agrifood waste, including vegetable crops that are damaged or unharvested, grain byproducts (e.g., canola or spent brewer’s grain), and dairy processing sidestreams such as leftover whey or cheese byproducts.
By analyzing patterns in existing data and combining them with machine learning models, the system will guide decisions on critical parameters: choice of microbial strain (e.g., yeast), fermenter type, temperature, pH, nutrient supplementation, and other variables that influence protein yield and quality.
This approach leverages the strengths of AI in handling large, complex datasets and identifying patterns that would be too difficult or time-consuming for humans to discern manually. By doing so, the tool aims to reduce the amount of trial-and-error experimentation needed, accelerate process development, and lower overall costs. The project will also make use of real-life fermentation records and practical industrial inputs to train and validate its predictive algorithms, ensuring that recommendations are both scientifically grounded and practically relevant.
Want to save for later? Click here.
The Influence on Global Food Systems
Although the project is early in its execution, the anticipated results center on making microbial protein production both technically robust and economically competitive with traditional protein sources. The research underscores that for this technology to truly influence global food systems, it must not remain a niche, high-cost alternative, but instead deliver cost-competitive solutions that appeal to food producers and consumers alike.
The CSIRO team notes that billions of tons of nutrient-rich material are currently lost each year. By transforming this waste into valuable protein ingredients, there are not only environmental benefits - such as reduced waste and greenhouse gas emissions - but also socioeconomic advantages, including the creation of new value chains and potential employment opportunities in the circular bioeconomy. This approach aligns with broader efforts to address food insecurity, especially in regions where access to protein is limited.
Importantly, the AI platform is expected to provide actionable insights for industry users. Instead of requiring deep expertise in fermentation science, producers could rely on data-driven recommendations to design processes tailored to their specific waste feedstocks. This could accelerate the adoption of sustainable fermentation practices and lower the entry barriers for small- and medium-sized enterprises seeking to engage in protein upcycling.
The integration of AI also has broader implications for food system innovation. By demonstrating how machine learning can optimize biological processes, the project contributes to the growing field of computational agriculture and digital bioeconomy. These innovations may lead to future developments beyond fermentation, such as AI-guided crop selection, precision agriculture, and supply chain forecasting - all of which could contribute to resilient and sustainable food systems.
Conclusion
The collaborative project between the University of Leeds and CSIRO represents a novel application of artificial intelligence in food science with the potential to tackle two critical global challenges: food waste and food security.
By developing an AI platform that recommends optimal fermentation conditions for transforming agrifood waste into microbial protein, the team seeks to make upcycled protein both economically viable and scalable. With funding from the Bezos Earth Fund’s AI for Climate and Nature Grand Challenge, this two-year initiative exemplifies how advanced technology can be responsibly harnessed to create environmental, economic, and social value. The success of this AI tool could catalyze widespread adoption of sustainable fermentation practices, reduce waste, and ultimately contribute to more resilient and equitable food systems worldwide.