How AI Is Reducing Food Waste

Who’s responsible for food waste?

For the large part, the answer is us, the consumers.

Back in 2021, the Waste and Resources Action Programme (WRAP), a UK charity, estimated that most thrown-away food (60%) can be attributed to households.

Annually, this level of spoilage and squandering amounts to 6.42 million tons of food going to waste in the UK alone. A separate 2019 report from the US Environmental Protection Agency estimated that the equivalent level of manufacturing food waste in the USA amounts to 40 million tons a year.

While these colossal figures are largely beyond the control of the food sector, the WRAP recommends two simple initiatives retailers and producers can adopt to encourage consumers not to waste their products. These include selling fruit and vegetables loose – so consumers can buy just the amount they need – and removing “best before” dates from packaging – so consumers aren’t instructed to dispose of perfectly edible food.

Since WRAP’s first report, these elementary proposals have been adopted by some UK food companies, and there is fledgling hope that the 60% figure will have shrunk when the charity next produces a report on UK food waste.

In the meantime, though, over in the world of food production, there are much more exciting technological initiatives happening to reduce the remaining 40% figure.

AI and food waste reduction

As estimated by WRAP, only 2% of food waste can be attributed to retail, 10% is down to hospitality, while a combined 28% can be attributed to farming and manufacturing (15% and 13%, respectively).

In other words, over a quarter of food waste in the UK occurs before the foods are even ready to eat.

That’s a figure Dini McGrath found hard to swallow.

“There's seven times more food waste that happens at a manufacturing level versus retail,” she told Technology Networks. “It is huge, but no one really knows about it. It's that sort of hidden figure.”

“And up until recently, food manufacturers haven’t been incentivized, other than for an operational cost perspective, to get to the bottom of it,” she added.

Having worked in the industry herself, at a fine gin distillery, McGrath understands the potential pitfalls of a production line. Supply chain issues. Demand forecasting errors. Poor equipment maintenance. Nine years at the distillery taught her there are innumerable ways waste can pile up.

Then something happened: powerful, next-generation artificial intelligence (AI) suddenly became accessible to businesses. For the first time, all the countless data in food factories didn’t seem so countless.

“There is information everywhere,” said McGrath, “but no human could actually come up with a solution without using technology to support them, just because of how complex the problem is.”

Empowered by this new technology, McGrath quickly co-founded Zest Solutions, “the first software empowering food manufacturers to take action against food waste”.

The company has since partnered with Nestlé. Speaking with Technology Networks, McGrath explained how Zest’s AI monitors the vast data of the European food giant’s production line to reduce waste.

Inside the factory

“Take a Kit Kat that is being made on a production line,” said McGrath. “You have a set amount of ingredients, all recorded digitally. You then have various different stages across the production line which will require melting of chocolate, cutting of wafers, various things to make sure you end up with your two or four sticked Kit Kat. At each stage you have food loss.”

“So, we use all of their different data sources, all of their weights on the production line and any other input and output data, to build a full picture of what is actually happening in that production line in real time, and use our proprietary AI algorithms to then deduce where these waste streams are going.”

Depending on the factory, McGrath says her team might install additional sensors, such as cameras, to capture missing data. In many cases, however, a factory’s existing sensors are adequate; the data they capture, fittingly, is simply being wasted.

“We do a factory tour so that we can map out [a production line] and actually build the right system, looking at their different production stages and where the data points match up,” McGrath said. “After that, we can literally use existing data.”

“There are cases where there are some gaps,” she added. “That is a case of applying cameras to get some visualization on a couple of areas, and then also a couple of weighing scales. So, in some cases data is sufficient, in other cases we are having to invest or get them to invest in a little bit of hardware.”

Once the number of sensors in a factory is sufficient, the data harvesting and processing begins in real time.

“Next we can then use these [data] to create insights that drive better operational changes, so we can identify hotspots of where there's loads of waste,” McGrath said.

“We can then suggest how they can optimize it,” she added. “We could say, ‘This is how many meals it would feed if you sent [Kit Kat food waste] here. If you sold it to a biscuit maker, then you could make X amount of revenue.’ We incentivize them to actually act if they can’t stop the waste occurring in the first place.”

Even these decisions of what to do with “wasted” food produce can now be made by AI. Sell on? Compost? Transform into an edible protein mixture? There’s an AI algorithm that can point the way, as Nicholas Watson, a professor of AI in food at the University of Leeds, explained.

AI and food waste: What to do with the refuse?

“Typically, if you want to reuse food waste, you can turn it into a liquid media, then you could ferment that with yeast,” Watson told Technology Networks last December. “Then the yeast will procreate and you'll get a microbial protein, which you can then process into food.”

“But there are lots and lots of different parameters. How do you treat the waste initially? How do you understand its varying composition? And then you go to fermentation and you've got things like temperature, the type of organisms you use, pH, how long you ferment it for. This is just a big problem,” he continued.

To supercharge this decision making process, Watson’s team began collaborating with Australia’s Commonwealth Scientific and Industrial Research Organisation on a new AI project.

“With AI, we can actually reduce the time and cost from maybe 25 experiments to
about 5 experiments,” he said. “We use some of our own data, but then we go to the literature and use tools to extract all the information and data in that literature to augment what we're doing.”

Just like Zest, Watson and his team are feeding their new technologies with existing, unused information to enable better decisions.

From optimizing factory floor workflows to deciding how best to ferment frittered food, it seems AI is finally making a meal out of wasted data.