As soon as the weather is good, many people rush to the baguettes for the perfect barbecue experience, in bad weather it can be a large piece of chocolate cake. Bakeries know these preferences. But predicting sales for each day is difficult for them. The environmental organization WWF estimates that around 1.7 million tons of baked goods end up in the trash in Germany every year. Apparently, human intelligence is not sufficient to predict exactly what will be eaten and when. Can it be the artificial one?

Up to twelve million tons of food are thrown away in Germany every year, almost seven million of which could be avoided. Approximately 34 percent of this waste is generated in the supply chain. This was the finding of a study by the Thünen Institute on behalf of the Federal Ministry of Food and Agriculture (BMEL).

Digital policy, regulation, artificial intelligence: the briefing on digitization

This waste is problematic on several levels. Wasted food is a waste of resources and also fuels climate change: According to a United Nations (UN) report, excess food is responsible for more than eight percent of CO2 emissions worldwide. For comparison: global air traffic before the corona pandemic for two percent.

Several start-ups have therefore developed software to prevent this. For example the Cologne company Foodforecast. It promises bakeries around 20 percent less food waste and up to five percent growth in sales – because the system also prevents goods from being sold out too early.

The team feeds internal data from the bakeries into the software, such as orders, sales and returns. External data such as the weather, public holidays or vacations are also important. The artificial intelligence (AI) tries to recognize patterns in the data and then spits out an order suggestion that the baker can follow.

There is a lot of interest in this, says Claudia Päffgen from Foodforecast. Rising raw material prices in particular are forcing bakeries to operate more efficiently. According to Päffgen, the 150 branches that have been using the Foodforecast software so far have been able to reduce their returns by 18 percent. In 2021, the start-up wants to have prevented 490 tons of food waste and thus saved 2.3 million kilograms of CO2. “We start with production and prevent food waste from occurring in the first place,” says Päffgen.

Another example is FoodTracks from Münster, which has equipped 1,500 branches with order optimization software – like the Härdtner bakery. Baker Nico Härdtner has been using the system for a number of years in his 60 branches. Since then he has recorded savings in the six-digit range. “It’s not fun for us to produce something that ends up in the bin,” he says.

Still, not everything about AI is new to him. “When I was a little kid, my mom wrote down what the weather was that day,” he says. The basic problem remains spontaneity. “Do you already know what you’re going to buy at the bakery tomorrow? The AI ​​doesn’t know that either.”

The Berlin start-up SPRK (pronounced Spark) takes a different approach. Here they want to redistribute excess food from the producers to dealers and needy customers – such as the Arche in Berlin, a Christian children’s and youth organization. To do this, the start-up receives data from the producer side, what is left over when and how much, as well as from the consumer side, what exactly is needed.

The AI ​​tries to recognize patterns in it in order to find a “match” as quickly as possible. With his startup, founder Alexander Piutti wants to create a platform for a functional circular economy. Not only those who lose weight benefit from this; they get the excess goods five to ten percent cheaper. The selling companies also save high disposal costs and share in the proceeds. According to Piutti, one kilogram of saved food avoids 2.5 kilograms of CO2 emissions on average.

Piutti’s ambitions should be of help to the federal government. In 2019, she launched the National Strategy to Reduce Food Waste. Their goal is to halve food waste in retail and among consumers by 2030. This could reduce greenhouse gas emissions by 9.5 percent.

Under Agriculture Minister Cem Özdemir (Greens), the strategy is to be further developed so that only half as much residue is left over in production and processing. Against the background of the war in Ukraine and the supply crisis, food waste must be tackled as a matter of urgency, according to the ministry. The authority is currently examining measures to facilitate the transfer of food.

The research project “REIF” of the Fraunhofer IGCV is about another use of AI – the dynamic price adjustment. Items with an earlier best-before date should be progressively cheaper than those that have a longer shelf life. “You know that. For three days, people probably all just bought the salad that keeps the longest. So why should I take the salad that expires in two days for the same price as the one that only goes bad in five days?” says Patrick Zimmermann.

The scientific employee is responsible for the project management of REIF. With the help of AI, digital price tags should automatically display the optimal price in the store. The idea is currently being tested in the Teegut market, says Zimmermann.

The project is pursuing other ideas on how AI can help reduce food waste. For example, platforms that are intended to network market participants or ways to make production more efficient. However, REIF is a research project funded by the Federal Ministry of Economics, emphasizes Zimmermann.

This means that no product is allowed to come out. It is only being researched to what extent AI can contribute to throwing away less. There are no valid results yet. Even with dynamic price adjustments, several price levels would first have to be tested in order to find out what appeals to customers.

Artificial intelligence can only learn if it is fed data – which the various companies first have to provide. In practice, there are also the cost issues. But from Zimmermann’s point of view, there is no other way to get better forecasts: “I always need AI when I can’t determine the algorithm through careful thought,” says Zimmermann.

Rainer Schramm, project manager of a dialogue forum at the German Agricultural Society (DLG), is also convinced that AI is a great lever for reducing food waste. According to the Thünen Institute, however, the biggest spenders are the consumers.

Private households are responsible for more than half (52 percent). Every year, 75 kilograms of food per person end up in the garbage. In the end, the problem remains a task for society as a whole. Schramm agrees: “It only works if everyone works together.”