Food Waste, Food Insecurity and AI

Press enter to search
Close search
Open Menu

Food Waste, Food Insecurity and AI

By Sivakumar Lakshmanan - 10/28/2020

Artificial intelligence has been used widely to address a huge variety of business challenges and opportunities. Where it hasn’t been deployed all that often is in the area of sustainability and general corporate responsibility.

AI can be used for activities like addressing food waste, which costs more than $1 trillion annually, and food insecurity, which now impacts as many as 40 million people in the US. There is knowledge from AI that can improve the way products travel from points of production to the points of consumption by identifying and re-directing food from channels that are more likely to result in spoilage and waste to channels that can alleviate food insecurity.

This, in turn can reduce the amount of food that winds up in landfills (about 30 million tons in the U.S., according to the EPA), as well as lowering fuel consumption, packaging and disposal costs, while providing a literal home for items on the verge of spoiling.

Yes, much of the food waste challenge occurs with consumers who are letting up to 40% of food products go to waste. But much of that waste, and the vast majority of the commercial waste is due to poor demand forecasting and supply chain management decisions.

The bottom line is that without effective ways to measure consumer demand, companies make decisions that don’t minimize food waste. In fact, some of the decisions actually exacerbate the problem by encouraging too large pack sizes or promotions that inevitably result in excess product.

To address these issues, food companies need to introduce specific solutions that ultimately match production or processing of a product with the consumption of that product. Think of a dairy company making a package of strawberry yogurt knowing that Mrs. Smith in a Chicago suburb is going to give it to her child next week. That insight aligns every transaction point in the supply and demand chain to ensure that just the right amount of strawberry yogurt is produced and distributed to meet the needs of the end consumer during that time period.

The key is that trading partners collaborate to accurately model the “perfect” supply chain, which includes consideration of the potential for waste with each unit produced.

Current best practice recommendations are to use artificial intelligence to generate forecasts from each level of demand, starting with the consumer and going back to the store, distribution center and production facility, then even further back to the ingredients suppliers and their sources.

Data can help route product from where it is likely to be discarded to areas of food insecurity prior to shipment from the production facility.

The consumer projection is based on the store point of sale forecast as a starting point, but also takes into consideration meal size, special events and other variables that impact at home consumption. Then the analytics and decision making structure can build a more holistic view of inventory need from store-level data on a range of factors, including minimum display quantities to safety stock minimums.

Demand forecasting is far more accurate when trading partners make use of AI technology that considers those external factors. For example, they can use AI to better manage the demand for essentials like water and Pop-Tarts prior to and immediately after a hurricane, and plan for bumper crops so nothing is left rotting in the fields.

In addition, the data can help route product from where it is likely to be discarded to areas of food insecurity prior to shipment from the production facility.

Based on past performance, food companies using AI-based integrated forecasting and replenishment solutions see dramatic reductions in spoilage and inventory. Other benchmark improvements include reductions of out-of-stocks and correlated increases in product service levels.

New AI and machine learning technologies can also optimize transportation so suppliers and retailers are alerted when too much product is being sent to one outlet and divert the shipment to a food bank in need.

Millions of tons of food in the U.S. and abroad continue to go to landfills while millions of people go to bed not knowing where their next meal is coming from. Using AI and machine learning, companies can address both of these issues simultaneously by optimizing the alignment of supply and demand, and better understanding how to adjust inputs for distribution channels to deliver outcomes that dramatically reduce food waste while directly addressing food insecurity.

Sivakumar Lakshmanan is COO, AI forecasting and supply chain at

More on Artificial Intelligence

While AI and machine learning are in widespread use across the CG industry, a level of confusion and uncertainty about their role in the enterprise and how to best deploy them is still prevalent. To uncover some of the ways they can have real value in retail execution, CGT hosted a virtual fireside chat with a pair of industry thought leaders.

During COVID there has been a boom in online shopping, particularly on safe “well-known brands.” However, managers of “invisible brands” might seek to connect by delivering information of full functionality on smart products to inspire future use and purchases.

F.A.I.R is part of the framework for Nestle’s data strategy as outlined in an opening keynote session delivered by CIO Filippo Catalano and Chief Data Officer Francesco Marzoni, at Analytics Unite 2020: The Summit for Retail and Consumer Brands.