A Google offshoot has created two new programs which make it easy for food producers, suppliers, and commercial kitchens to route unneeded food to food banks that need it.
Known as Project Delta, the machine learning programs take into account thousands of different calculations and variables, the things it would take a dedicated team of organizers to manage, to ensure the food is going where it is needed most, where it’s most likely to get eaten, and other priorities.
Food waste is what sports commentators would describe as “a good problem to have,” as it inherently suggests there’s enough to go around. The problem is that it’s not always going to where there are hungry people.
The issues are mostly found in the supply chain: through the interaction of buyers, trying to imagine how many units they need, and sellers who are trying to imagine how many they can sell.
Producers, like sellers, try and make as much as they think they can find buyers for, while the final stage owner of food—supermarkets, restaurants, or hotels, often have too many processes to worry about to consider how best to send food further down the line.
The efficient distribution of food is an extremely difficult job to take on all the way down the supply chain, and so it’s perhaps no surprise that 30-40% of food in the U.S. is wasted.
“There’s no simple way for food suppliers to let food banks know what they have available, or for food banks or pantries to communicate what they need,” writes Adele Peters for Fast Company, covering the Google innovation.
MORE: The 2020 Nobel Peace Prize Goes To The World’s Largest Hunger Program
The idea is being hammered out at a Google offshoot called X, which styles itself as “The Moonshot Factory,” and focuses on providing super innovative solutions to make the world a better place.
A better way to bank
Emily Ma, writing for X’s blog on the two years of development and testing for Project Delta, explains they set out “to create a smarter food system — one that knows where the food is, what state it’s in, and where best to direct it to ensure it doesn’t end up in a landfill and instead goes to the people who need it most.”
Some of the problems those working on the problem at Google faced were things like a lack of industry or cross-industry standard for how food suppliers communicate what they have or what they need to move most, or in terms of food banks, what they need and what people aren’t eating.
She jokes that there isn’t even an industry standard designation for the state of Texas, and that during her and her team’s preliminary research, they found 27 different words in organizational data i.e. TX, Texas, Tx, etc.
Working with the Southwest Produce Cooperative’s (SWP) food banks in states like Arizona, Ma and her team first built their machine learning prototype, which in place of phone calls, emails, site visits, and paper records—the normal ways the SWP food bankers coordinate shipments— uploads all records relating to supply and demand into the algorithmic bot which details what should go where, and when.
Next Emily Ma and the Moonshot Factory team went to Kroger to see if they could improve logistics at Feeding America, the country’s largest domestic hunger-relief organization.
Their program insights into the operation of Kroger’s delis, in which meat and other products are typically recycled due to health and safety concerns, allowed them to open up opportunities to give millions of additional meals to communities that need it.
Finally, in a more impressive display of machine learning, cameras installed next to waste bins in Google-facility kitchens were able to collect twice as much information about food waste as the manual by-hand logs made by chefs which took about 30-60 minutes to complete.
The machine learning can identify trends, such as larger amounts of a particular food being wasted, as well as make recommendations for dishes and ways of recycling that can reduce food waste in commercial kitchens.
SHARE This Fascinating Story With Friends on Social Media…