Leveraging AI on the Road to Net Zero

Last summer’s severe heat waves around the world broke through the global consciousness and highlighted the possibility that climate change had reached dangerous new levels. Other dominant headlines trumpeted the public launch of generative artificial intelligence (AI)such as ChatGPT as it promised untold new wonders for humankind.

So is there hope for the future in putting the two trends together? Why not apply the exponentially growing capabilities of AI to solving the problem of an overheating planet? That seems to be a growing bet, and the food industry finds itself in the center of the action. As a business with one of the biggest impacts on Earth’s sustainability, food also controls some of the most important potential mitigators of that impact.

AI can be an incredibly significant tool for sustainability [within] the food industry when it’s used properly.

“AI can be an incredibly significant tool for sustainability [within] the food industry when it’s used properly,” says Jonathan Darling, the leader of automation for consumer packaged goods customers for Schneider Electric, an industrial equipment manufacturer.

Indeed, industry players now are applying AI to advance along the path to net zero throughout the entire supply chain for ingredients, foods, and beverages. Their initiatives after the farm gate stretch through transportation, processing and manufacturing, logistics, merchandising, inventory control, and retail waste.

As it produces more energy-saving efficiencies along the food supply chain, AI also is providing the industry with potentially robust new ways to cope with stiffer global regulation and governments’ increased coordination aimed at cutting carbon dioxide emissions, including life cycle analysis of the sustainability impact of individual products. The rules govern what are known as Scope 1 emissions, which cover those a company makes directly; Scope 2 emissions, covering indirect outputs, such as those by electricity being generated on a company’s behalf; and even Scope 3 emissions, which comprise all emissions associated with suppliers and customers up and down a company’s value chain.

Food companies and their suppliers are finding these pressures especially strong in Europe. European and British companies “are very much looking at helping with the environmental impact,” says Sam Vise, CEO of Optimum Retailing, which offers an inventory management system for food retailers. “But in North America, it more often comes down to profit: Companies want to reduce waste and increase revenue and improve the customer experience.”

Indeed, it’s fortunate for the food industry that the pursuit of greater sustainability using AI mostly overlaps with its long-standing business goal of enhancing profitability. The ever-improving technology likely is the best one to come along to advance the food business in two of the three elements of the “triple bottom line” of profit, planet, and people.

Business conversation drives things first with AI, with sustainability outcomes a nice extra benefit.

“Business conversation drives things first with AI, with sustainability outcomes a nice extra benefit,” says Maria Pearman, principal, food and beverage, with GHJ, a consulting firm. “A lot of these technologies are allowing entities to operate in a more efficient manner, which means better use of dollars, time, and resources. And all those things add up to a lighter environmental footprint.”

Consider Mondelēz International. The worldwide snack giant has been applying the principles of AI to its production processes “for a while,” says Min Ling Chan, senior director of global engineering for digital, automation, and packaging. But the past few years “have seen an increase of pilots and adoption,” with enhanced AI creating advances to review “toward improving production processes, reducing costs, enhancing quality control, and achieving greater efficiency.” All of that, of course, contributes to greater profitability as well.

“You hear more about sustainability, but profit didn’t go away,” says James Newman, the head of product and portfolio marketing for Augury, which leverages AI-driven insights to help clients improve manufacturing processes. “The conversation is now about how to make them symbiotic. And with AI, we can actually do more to not have sustainability and profit objectives fight against each other, something we couldn’t have done even several years ago.”

That doesn’t mean the capabilities of AI are an easy sell: 71% of manufacturers recently surveyed by Augury “still think sustainability targets hurt or don’t have any impact on [improvements toward] their production goals,” Newman says. “Forever, food manufacturing has been about throughput, quality, that kind of thing. People have thought of sustainability as a side thing that you do because people want you to. But what if it was the same thing: Get the machines running better, and achieve sustainability. It doesn’t have to be a zero-sum game.”

In fact, the food industry’s investigation to see how AI can advance sustainability goals is tempered by the understanding that this technology may not prove to be as game-changing in the food industry as in many other businesses nor even create as much impact on food manufacturing and retailing as AI does in other aspects of the food business, such as agriculture and product development.

AI is mainly just an enabler of incremental improvement in the accuracy of what we’re already doing.

“AI is mainly just an enabler of incremental improvement in the accuracy of what we’re already doing,” says Greg Wilson, vice president of sales and field strategy with RELEX Solutions, a supply chain optimizer for the grocery business.

Adding Efficiencies, Cutting Waste

It’s already clear that AI has the potential to play a major role in adding operational efficiencies that can reduce food manufacturing’s environmental footprint, through a symphony of big and little steps that leverage AI techniques in transportation, logistics, production, and retailing to cut energy usage and thereby emissions, to use manufacturing resources more efficiently, and, at the end of the food chain, greatly reduce merchandising waste.

For example, in just six months of working with one North American manufacturer, Newman reports that by Augury’s calculation, the company avoided 4,500 tons of carbon dioxide emissions “just by keeping their machines running better” relying on AI technology. That same exercise cut waste by 45% on one processing line and reduced energy use by about 5%.

“This is not a theoretical exercise,” Newman says. “That’s where the transition on sustainability has to happen. Using solar and other alternative energies may be a long way off for a lot of manufacturing assets, but AI can help us do tactical things on the shop floor that make a big difference already because incremental things add up.”

Mondelēz has enhanced its process control by employing machine vision systems that utilize image recognition, a form of AI. “This technology provides real-time data about the status of our processes and aids in minimizing production waste and maintaining high quality standards,” Chan says. The company also is using AI-powered robots in packaging for tasks such as sorting, in systems that “collaborate with human workers to enhance both efficiency and safety on the production floor.”

Catena Solutions, a supply chain consultancy, has worked with another large CPG manufacturer that is implementing AI, robotics, and other forms of automation in its plants “to do heat mapping and heat thermal scanning of their lines to make sure that when something comes to near the end of life it doesn’t break before they need to replace it,” says Geoff Coltman, vice president, client engagement. “That helps sustainability because you’re reducing waste in your product.”

Consider, for instance, how AI systems are helping manufacturers improve their use of extruders. Downtime for extruders is one of the biggest time- and resource-wasters in food factories, and AI can “make it so humans are more efficient in making decisions,” Augury’s Newman explains, “to co-pilot with people, not replace them, based on synthesized information. AI helps us combine information to present to people the things they need to pay attention to, which on extruders are machine health and peak performance. Does it have bearing or lubrication issues? AI can synthesize millions of data points an hour on that machine, but all the end user sees is when there’s a problem and they need to take action.”

In general in manufacturing, Darling says, one of the most helpful aspects of AI is that it drives bias out of decisions where humans often introduce it. “AI does that by contextualizing the data, allowing for better decisions to be made,” he says. For instance, Schneider has worked with a brewer that uses computer vision, machine learning, and AI to analyze foam levels in beer during the bottling process to make sure each bottle meets quality standards, “drastically reducing waste and bad product in the field.” Candy makers and bakers are using AI to set cooking times and temperatures to optimize taste and texture and reduce waste from human error.

Reducing Retail Food Loss

Moving up the supply chain from the manufacturing plant to the supermarket, AI is helping address another major challenge: food lost to spoilage. Leveraging AI can automate essential tasks such as shelf-life monitoring and expiration-date tracking, playing a pivotal role in minimizing waste. At the same time, the improved accuracy in inventories resulting from AI implementation helps retailers to make data-driven decisions, taking the human bias element out of it, similar to what happens with altering machine settings on the factory floor. That ensures grocery products are merchandised and purchased optimally, curbing unnecessary waste. AI also is helping cut waste by identifying daily, weekly, monthly, and seasonal fluctuations, buying patterns, and supply chain dynamics.

“Demand planning is a significant portion of AI and predictive analytics because if you go backward through the production chain, a piece of sustainability is making sure foods and beverages don’t go to waste,meeting demand but not oversupplying demand, and making sure what you need to meet demand isn’t wasted from an ingredient standpoint,” Coltman says.

When it comes to the sustainability impacts of AI, “the biggest opportunity is to take waste out of the system,” says Sivakumar Lakshmanan, head of software products for Zebra Technologies, a provider of inventory technology systems that acquired an AI provider a couple of years ago. “You can’t talk net zero if you’re throwing stuff out. There is a need for considered effort across the food value chain to reduce this waste, and AI definitely is going to play a bigger role, partly by figuring out exactly what the customer needs tomorrow.

“The problem is that all retailers need variety and speed. But we also need climate change to be addressed,” says Lakshmanan. “We have to balance these conflicting priorities, and AI comes into the picture and helps you make better decisions.”

Using Zebra, for instance, retailers can select how much of each kind of bread SKU they want based on AI analysis of “what are the driving factors and market demand and larger trends in the market,” Lakshmanan says. They include, for instance, whether white bread might sell better on the weekends for the beginning of the school week and what effects retailers can expect from promotional pricing or a marketing campaign.

AI also helps retailers better comprehend exactly what they’ve got on store shelves so they can manage it more efficiently. Optimum Retailing, for instance, recently teamed with Amazon Web Services on development of an AI-based tool that leverages image recognition to provide precise real-time data to retailers and manufacturers about actual inventories in the store instead of what datasets collectively say should be there. It has the potential to tremendously boost accuracy and save countless hours of labor in a worker-tight industry.

So, AI trains a computer about how a box of Kellogg’s Corn Flakes or a can of Del Monte Creamed Corn looks on a shelf “when we don’t know exactly where we’re going to see that object in different environments,” Vise says. “You have to give the computer all the possible scenarios, but AI models are very efficient in processing immense amounts of data.

“AI helps us see patterns in when products are put on a shelf, when they are purchased, and what portions are waste,” Vise says. “You can see that shopping habits are different on Sunday morning than Sunday evening. So if you’re going to put something on display, maybe a certain fruit should be there, or maybe family packs of yogurt instead of individual tubs as people buy stuff for the week.”

The biggest opportunities for cutting waste at the retail level lie, of course, in fresh foods, especially produce. In addition to waste, produce is where most safety scares develop, posing yet another reason the industry wants to enlist AI to do a better job of monitoring and managing supplies of fruits and vegetables.

For example, Wilson says, bell peppers with two weeks of indicated shelf life may not be the same as tomatoes with two weeks of shelf life. RELEX’s AI system interprets customer behavior to figure out the differences and recommend responses to retailers, and does so much more efficiently than traditional statistical analysis.

“If customers stop buying tomatoes after a week, but not peppers, something is going on,” Wilson says. “Maybe the skin starts wrinkling a bit, or they might be softer. What we can do with the [retail] customer is say, ‘If the tomatoes reach a week, let’s move those into in-store production and use them to make salsa or tomato sauce.”

In fact, Wilson says, RELEX recently calculated the impact of its AI on inventory control for its fresh produce retailers and found 5% to 10% improvements in out-of-stock situations, the ability to reduce inventories by from 5% to 20%, and spoilage reductions ranging from 10% to as high as 40%. And, feeding back to the important “profit” element of the double bottom line, these retailers consequently were able to reduce their trucking costs for produce by 5% to 10%.

AI also is being enlisted to help Farmer’s Fridge, a 10-year-old fresh food vending company that tries to take advantage of every technology-based increment not only to insure against spoilage but also to put the right products in the right places—airports, college campuses, and hundreds of other locations—for the right consumers. All of that prevents waste. Advanced demand planning “allows us to predict what our aggregate demand is going to be, say, a month from now,” founder and CEO Luke Saunders told Kiosk Marketplace. “That allows us to plan our labor, [and to] plan our purchasing, which allows us to be much more efficient on the production side.”

Planet FWD is a four-year-old company that is compiling data and using AI to score the “cradle-to-grave” environmental impact of thousands of individual food products for companies that want to leverage the information, including Compass Group, Blue Apron, and Patagonia Provisions. Clients use the product ratings and Planet FWD’s advice to influence Scope 3 emissions by suppliers, to communicate marketing messages on their websites and elsewhere, and to improve their sustainability compliance.

“We’re moving from just reporting to helping them reduce” environmental impact, says Planet FWD CEO Julia Collins. “They can see not only their historical and current emissions but also a projection of their emissions into the future based on the trends in their business and how, by taking certain steps, they can accelerate their path toward net zero.”

Planet FWD’s calculations even take into account the environmental effects of how consumers use an individual food product. “For most food products,” Collins notes, “end-of-life uses aren’t trivial. With a box of tea, for instance, heating up the water for it is one of the highest-impact aspects of the life cycle of the product.”

Limits and Risks

AI won’t be a panacea for getting food systems to net zero, however. One factor likely to hold back adoption of AI-based solutions is that the industry typically is a late adopter of technology such as robotic automation, compared with manufacturers in other major verticals, including pharmaceuticals, aerospace, and automotive, says Schneider Electric’s Darling. In a recent survey of food manufacturers, for example, 52% said it is very important for their operations software to include AI and machine learning capabilities, but only 34% said they were very prepared to leverage such capabilities.

There’s another broad caveat: For all manufactures, not just those that make foods and beverages, generative AI, the shiny new object in the technology realm, isn’t yet ready to contribute much on the factory floor. “ChatGPT isn’t necessarily supporting manufacturing processes yet,” notes Catena’s Coltman.

Then there’s this: Do we really want it to work? The news about generative AI prompted apocalyptic warnings from some of the originators of the technology about the threat of an actual AI takeover of, well, the human race, in the stuff of science fiction movies. It doesn’t seem the food industry would be the conduit for such a scenario, but AI does pose risks, such as security and privacy issues.

“That goes hand in hand with the use of AI,” GHJ’s Pearman says. “My hunch is that risk is always going to outpace the preventive technology that is embedded. It’s always going to be something we’re chasing, and there will always be the balance of doing the best we can and not letting negative effects outweigh positive effects.”ft