As with so many things, the pace of change in the food industry is increasing rapidly and agriculture is not immune to the changes of the digital age. Technological innovations can transform every link in the food chain, from seed to fork. Digital technologies are transforming the business – you have robots, augmented reality, virtual reality, 3D printers, data analytics and sensors, drones, and cloud computing – but they all have one thing in common: artificial intelligence (AI). AI is the secret sauce behind them all. 

AI refers to the collection of data from sensors and its conversion to comprehensible information. AI machines can mimic human cognitive functions such as learning and problem solving, and have the potential to interpret information more efficiently than humans, reducing their need to be involved.

An exciting example in agriculture is machine vision, whereby computers process visual data collected via unmanned aerial vehicles (drones), satellites or even smart phones and provide the farmer with useful information. 

For example, Labby Inc is a start-up that has spun out of one of America’s leading universities, MIT, and is using AI to analyse the data from milk sensors to detect changes in milk fat, protein and even somatic cell counts. 

Cainthus processes camera images of cows and has developed algorithms to identify animal behaviour, feed availability and productivity in dairy herds. In this way, AI can potentially aid with better decision-making, it can be self-learning, and can progress beyond human abilities. But as we have seen in other industries and settings, such as hospitals, its real power comes from the potential to make people better at their jobs, not to replace them. The use of AI to advance food production is accelerating as the world progresses post-Covid-19 and expectations of speed and efficiency, as well as sustainability, are ever-increasing, even as the global population grows rapidly. The rest of this article looks at some examples of AI in the food industry and how it has accelerated growth, or even changed the way a company operates. 

 

Processing 

The processing of food is a labour-intensive business, but one where AI can maximise output and reduce waste by replacing people on the line whose jobs are to identify items unsuitable for processing. Decision-making of this type at speed requires the senses of sight, smell, and the ability to adapt to changing circumstances. AI brings even more to the table through augmented vision, analysing data streams unavailable through human senses. 

TOMRA, a manufacturer of innovative sensor-based food sorting systems, is incorporating AI technology into its processes to detect abnormalities in vegetables and fruits, including sensor-based sorting machines, detecting and removing any types of foreign materials from its lines of produce, reacting to changes in moisture levels, colours, smells, and size. TOMRA’s primary focus is reducing food waste, for example they claim the recovery in potato processing of ‘5-10% of produce through higher yields and better utilisation’ equivalent to ‘25,000 trucks of potatoes per year’. TOMRA is currently developing meat-processing applications. 

Another example is how Japan’s food-processing company Kewpie uses Google’s Tensorflow AI for the detection of defective ingredients during processing. It was originally used for the sorting of foods, and gradually developed into an anomaly detector, which could then be used for unsupervised learning, saving both a large amount of time and money. While focused on diced potatoes currently, the company plans to, “expand to eggs, grains, and so many others.” 

Dutch company Qcify offers automated quality-control and optical-monitoring solutions for the food-processing industry. It uses machine vision systems to classify almonds, pistachios and other nuts. It claims to be able to identify quality at twice the speed of a human operator and more importantly to automatically remove impurities or foreign objects, and automatically generate quality reports. 

A myriad of other agri-tech start-ups are focussed on using AI to detect early warning signs of poor health in crops that may have otherwise been overlooked, which can further reduce the waste in food production, in addition to increasing transparency. 

 

Supply chain efficiencies 

The food-ordering and delivery app, Uber Eats, is now incorporating AI to make ‘recommendations for restaurants and menu items’ and to ‘optimise deliveries’, as well as looking into the use of drones. It uses Michelangelo, Uber’s machine-learning platform, for various tasks. For example, this can predict meal estimated time of delivery (ETD) to reduce waste and improve efficiency throughout its delivery process. While this application is post-food production, once food has been produced, there are many more ways to implement this up and down the chain. 

Covid-19 has accelerated the applications of technology to replace human labour and while smart-device food apps, drone and robot delivery, and driverless vehicles all provide new ways to get information and food to the consumer, all of them depend upon AI. 

Innovative uses of AI are crucial in moving towards reducing the quantity of food wasted to feed the growing world population as efficiently as possible, as well as meeting increasingly specific consumer demands and expectations. 

Supply-chain forecasting company Shelf Engine employs AI to remove human error while working with perishable foods from the purchasing function and making more informed decisions about order sizes and types in hundreds of US stores. It has saved thousands of dollars in food waste. 

Wasteless is a machine-learning and real-time tracking solution that allows retailers to use dynamic pricing to discount produce before it goes past its sell-by date. 

 

Overcoming the odds

Alongside all the positive aspects of AI, some see it as a technology with the goal of taking over human jobs, and that creates controversy. The fear of the unknown is creating a pushback against the use of AI in many businesses. Additionally, AI requires skilled IT professionals, who are in high demand and difficult to recruit. Clearly, there are costs to retraining programmers to adapt to the change in skills required. Finally, the costs of implementing and maintaining AI are very high, which may limit the opportunities for smaller or start-up businesses to compete with already-established larger ones. Downsides such as these could possibly slow down the speed with which AI transforms food production, but given the absolute power unleashed by AI in a post-pandemic food world, it is unlikely to be more than a speed bump on its eventual universal acceptance.

 

Food safety 

• Reducing the presence of pathogens and detecting toxins in food production are key avenues for AI. The Luminous Group, a UK software firm, is developing AI to help prevent outbreaks of pathogens in food-manufacturing plants, limiting consumer illness or recalls. 

• Additionally, AI offers the opportunity to increase traceability and consequently, consumer confidence. For example, Remark Holdings, a subsidiary of data-intelligence platform, KanKan, uses AI-enabled cameras in Shanghai’s municipal health agency to check that workers are complying with the safety regulations. This algorithm-based machine-learning technology includes facial and object recognition, and ‘sets the foundation (...) to potentially triple [their] business with the city of Shanghai’ according to Remark Holdings chair and CEO, Kai-Shing Tao. More recently, the company added improved facial recognition abilities to account for the mandatory use of a mask. It also introduced a new body temperature detection solution, in response to the effects of Covid-19. By identifying instances of increased body temperatures, this new technology could help in the early detection of a Covid-19 case. The ever-evolving project shows an ability to constantly grow and develop, a flexibility that is essential today in the world of technology. 

• Additionally, Japanese technology company Fujitsu has developed an AI-based model that can monitor hand washing in food kitchens following strict regulations set by the Japanese health ministry. This technology will reduce the need for visual checks (critical during Covid-19) where food safety must be increased. 

• Next-generation sequencing (NGS) is also another attempt made by food safety associations to increase the accuracy and speed with which any threats to food safety are identified and addressed in a production chain. AI can be used for ‘cleaning in place’ projects, which aim to use AI to clean production systems more economically using more environmentally friendly methods. 

• In Germany, a project by Industrial Community Research strives to ‘develop a self-learning automation system for resource-efficient cleaning processes’. This develops a cleaning process without a need for the disassembly of equipment. This could cut labour costs and time spent on it, as well as increasing the safety of food production by removing the potential for human errors. The University of Nottingham has also been working to construct a self-optimising clean-in-place system, which uses AI ‘to monitor the amount of food and microbial debris in the equipment’.

 

About Aidan Connolly

Aidan J Connolly is a contributor to Forbes. He teaches at three Agri-Food MBA programmes, and is adjunct agri-business professor at China Agricultural University. He is president of AgriTech Capital, advising on innovations and technology investments. His most recent book The Future of Agriculture is a free download at: www.agritechcapital.com/books