Using Artificial Intelligence to make more sustainable behaviour easy for people.

Artificial intelligence will make social and environmental-friendly choices much easier.

The majority of German consumers are willing to shop more sustainably: 68 per cent are willing to pay more for a product that is proven not to harm the environment, according to a survey of 2,500 consumers conducted by the auditing and consulting firm Ernst & Young (EY) in early 2020. Consumers are increasingly researching products on offer themselves. 46 per cent of the respondents said that materials and ingredients are decisive factors in their purchase decision. "Up until now, researching sustainable product alternatives costs consumers a great deal of their time," Ruben Korenke, project manager of Green Consumption Assistant (GCA) told RESET.org. “The advice provided is often too disconnected, geographically and topically. So when it comes to making a purchase decision, the information may no longer be relevant." This means that even those with a high level of environmental awareness will consume less sustainably.

Enabling an informed purchase decision

This is exactly where the GCA comes into play: "We want to make sustainable behaviour and buying online as easy as possible," says Korenke, summarising the goal of the cooperation between TU Berlin, Beuth University of Technology and Ecosia GmbH, the sustainable Berlin-based search engine. “We know that many people would like to pursue a more sustainable lifestyle, but are not currently doing so. We want to make it easier for these consumers to choose sustainable alternatives.” In order to achieve this, the smart assistant functions as a browser extension in search engines such as Google and shopping platforms such as Amazon.

When a consumer searches for a product, the GCA will inform him or her directly on the chosen online platform about the CO2 emissions, manufacturing conditions and the ecological footprint of the desired item. It will also provide links to more sustainable alternatives that have been produced in an environmentally sustainable and socially just way. "Once you have installed the GCA, it will suggest sustainable alternatives," says Korenke. "For example, if you search for a smartphone on Amazon, it will suggest recognised sustainable phone products such as a Fairphone." It will also show links that provide information on the benefits of keeping your smartphone for a longer period as well as tips on repair, rental and sharing options that can save you the expense of buying a new product. Because "sometimes the most sustainable option is not to purchase at all," adds Korenke, who, in addition to heading the GCA project, is working on nudging for sustainable consumption as part of his doctorate. This relates to how the situations in which we make decisions affect the environmental impact of those decisions.

Artificial intelligence makes it possible

The recommendations provided by the GCA will be made available via a database with sustainability information about products and services, which was created in October 2020 at the Einstein Center Digital Future in Berlin. However, this is precisely where one of the greatest challenges of the sustainability project lies: Since this database is new, it must first be built up. "AI can help us to merge sustainability data on many products from different data sources," says Korenke. He and his team rely on the possibilities of machine learning. The TU research group aims to create a basis for the consumer assistant based on these existing data sets. During the first phase, the focus will be on product categories and average values, data on individual products will then follow. This is not the only area where artificial intelligence is beneficial: “It can help us to find the right time and place to provide sustainability recommendations. For example, our models learn what kind of search would be appropriate for a smartphone recommendation," says Korenke. "However, despite all this, it is important that we use AI moderately - the energy consumed by training AI models can be considerable and we should bear this in mind."

AI: Tech fix or eco problem?

From AI-controlled clean wind farms and sustainable supply chains to automated evaluation of satellite images and monitoring aid measures in crisis regions - the list of sustainable AI use cases is growing steadily. According to the AI map, in Germany alone there are around 150 projects that use AI to help solve pressing social and ecological problems. In some places the technology is hyped as a "game changer for climate change and the environment". According to a study, the use of AI levers could reduce global greenhouse gas emissions by four per cent by 2030. That would amount to a total of 2.4 gigatonnes of CO2 emissions - this corresponds to the combined forecasted emissions of Australia, Canada and Japan in 2030. The potential is undoubtedly great. However, the volume of energy and resources AI systems such as deep learning consume for computing processes should not be overlooked. According to a study by the University of Massachusetts, training a powerful and modern artificial neural network used for speech recognition causes 0.65 tonnes of CO2 emissions. That is five times as much CO2 as a car emits during its entire service life. Or to put it another way: It is the equivalent of a return flight from Berlin to Madrid.

“The energy and resources that AI consumes has seldom been adequately reflected upon. I would like to see a greater emphasis placed on this,” says Korenke. The use of AI in the sustainability sector only makes sense if, for example, the energy consumption of the training and use phase is lower than the energy and resource savings that are ultimately achieved through the process. He then advocates “using AI with enthusiasm to make it easier for people to enjoy a more sustainable lifestyle.” A good example of this is the Stena Fuel Pilot project, in which the ship route between Kiel and Gothenburg was optimised with the help of AI. Although learning the AI ​​would have been expensive and energy-intensive, if the pilot project was upscaled and the AI ​​tool was transferred to the entire fleet in the long term, each ferry would save two to three per cent fuel on each crossing. This in turn ensures a positive ecological balance in the long term.

Beta test phase at Ecosia

The GCA project, which has been in the beta test phase since February 2021, is also hoping for a lasting positive effect: Since the beta was launched, users have been able to test the browser extension on the Ecosia search engine and participate in its development. In addition, Ecosia's map service highlights places that offer sustainable products and services, such as vegetarian restaurants, open workshops, rental stations and second-hand shops. "Together with Ecosia founder Christian Kroll, we were able to launch a truly special consortium of machine learning expertise, sustainability knowledge and experience in the development of sustainable digital products," enthuses Korenke about the collaboration with the sustainable search engine. The test phase, which currently only works for online searches for smartphones, is to be expanded step by step to other product categories. "Naturally it would be great if at some stage we could also include all information on social and ecological sustainability, for example the influence on biodiversity or human rights compliance in production," says Korenke, explaining his vision for the project, which is funded by the Federal Ministry for the Environment, Nature Conservation and Nuclear Safety (BMU) as part of the AI Lighthouses for Environment, Climate, Nature and Resources initiative.

Berlin: A city for networking and inspiration

That is not his only vision: If successfully implemented, it will allow millions of consumers to interact with the GCA, developed in Berlin, via the Ecosia search engine every day - in over 30 countries. In addition, the database will be published as open data and open source so that other search engines and content providers can also access it. The team from Berlin hope that this approach will lay the foundation for new research and development projects. Furthermore, new digital business models and start-ups are expected to emerge from the GCA. Korenke is optimistic, especially given the benefits Berlin offers tech companies based in the city. “Our company history is closely linked to the networking system in the R&D sector in Berlin,” says the project manager. “In my opinion, such ideas tend to succeed where there are opportunities for networking between the different partners - in this respect Berlin is certainly unique in Germany."