1. General
Category
SDG 3: Good Health and Well-being
SDG 9: Industry, Innovation and Infrastructure
SDG 11: Sustainable Cities and Communities
SDG 12: Responsible Consumption and Production
SDG 13: Climate Action
Category
Food Processing & Sales
2. Project Details
Company or Institution
Linz Center of Mechatronics (Dr. Johannes Klinglmayr), ETH Zurch (Prof. Dirk Helbing), University of Leeds (Dr. Evangelos Pournaras)
Project
ASSET: A Value-sensitive AI Approach to Empower Sustainable Consumption
General description of the AI solution
Creating more sustainable consumption patterns turns out to be imperative for mitigating climate change and supporting a more viable future of our society. The ASSET EU project has introduced a novel and value-sensitive AI solution with which consumers make product choices, e.g. groceries, that meet their sustainability preferences, for instance, fair trade or vegetarian products.
The core of this AI solution is an innovative product rating systems implemented as a smart phone app. The consumer detects all nearby products with its smartphone and the app calculates a personalized score from 1-10 measuring to what extent each product satisfies the sustainability preferences of a consumer.
This score is a game changer – it summarizes in smart way an enormous amount of information about product characteristics and consumer preferences to allow simple and fast product comparisons leading to more sustainable purchase choices seamlessly integrated in the shopping process, i.e. during a visit in a retail shop.
The design of the ASSET product rating system is value-sensitive: distributed privacy-preserving computations empower an open-source content-based recommender algorithm to calculate efficiently explainable product ratings that capture for first time such a broad spectrum of human and sustainability values with ease. The AI algorithm reasons about the sustainability of products using a prominent sustainability ontology co-created with sustainability organizations, experts and `wisdom of the crowd’. With such novel value-sensitive design approach, the ASSET AI product rating is more trustworthy, transparent and accountable and can be used as a digital blueprint to scale up sustainable consumption at a global scale with the active citizens’ engagement.
Real-world field experiments in two supermarkets in two different countries confirm higher sustainability awareness and a bottom-up behavioral shift towards more sustainable consumption. These results encourage novel business models for retailers and producers, ethically aligned with consumer preferences and with higher sustainability.
Website
Organisation
University of Leeds
3. Aspects
Excellence and Scientific Quality: Please detail the improvements made by the nominee or the nominees’ team or yourself if your applying for the award, and why they have been a success.
The product rating system is an advanced AI-content-based recommender system with several novelties:
1. Privacy-preserving personalization. To transparently calculate products rating, summarized product characteristics on sustainability are transferred to consumers’ smart phones for processing locally together with consumers’ sustainability preferences. No sensitive data are shared to third parties.
2. Explainable product ratings. Consumers can easily explore which of their sustainability preferences and product characteristics contribute to product ratings and how. This functionality is also transparent, designed to operate locally on consumers’ smart phones.
3. Reasoning based on experts’ knowledge and wisdom of the crowd. High-quality sustainability information about products is collected, processed and structured in a novel sustainability knowledge-base. It combines official reports from environmental/health organizations, experts’ feedback and citizens’ involvement via hackathon events.
4. Practical/compatible to real-world shopping processes. The product rating system seamlessly integrates into an augmented shopping experience: products nearby to consumers are automatically identified by their smart phones using wireless communication technology. This approach revolutionizes the ease of comparing products in contrast to barcode scanning technologies.
The product rating system was demonstrated at TRL-7 in two real-world supermarkets (Estonia and Austria) during May-November 2018. These field tests clearly showed a causal relationship of the product rating with the shift of consumers to buy more sustainable products, even if these products may be more expensive. This published work has received extensive coverage in media including news articles, interviews as well as appearance in live science and technology TV programs. The core AI-based product rating system is open-source for the broader community and system applicability. The project has received the R&D Prize 2018 on innovation by the state of Upper Austria (https://www.land-oberoesterreich.gv.at/209106.htm). Currently it is exhibited at the Vienna Biennale for Change 2021.
References
Key publication
https://royalsocietypublishing.org/doi/pdf/10.1098/rsos.201418
News media
https://science.orf.at/stories/3204123/
https://www.blick.ch/wirtschaft/psychologie-wie-man-mit-apps-nachhaltigeren-konsum-foerdern-kann-id16294311.html
https://www.amna.gr/home/article/522381/-Protoporiaki-efarmogi-kinitou-boitha-tous-christes-sti-biosimi-katanalosi-sta-soupermarket-
TV coverage:
https://www.youtube.com/watch?v=RCyFoMqGUq8
https://www.youtube.com/watch?v=JNXKMznE58E
Scaling of impact to SDGs: Please detail how many citizens/communities and/or researchers/businesses this has had or can have a positive impact on, including particular groups where applicable and to what extent.
The ASSET project has a significant impact at different levels:
1. Consumers. Our field test results show that consumers significantly change purchasing habits and repeatedly use the AI tool when shopping with their money in an Estonian and Austrian supermarket. A transparent product rating with a value-sensitive AI approach reduces information complexity and empowers change in citizens’ shopping behavior. Our approach demonstrates an intrinsic consumers’ will to encounter SDGs (addressing SDG 12-13). Beyond the field tests, 90% of consumers prefer labeled than unlabeled products (Ipsos and London Economics Consortium 2013). With 65% of citizens having a mobile device with internet access (STATISTIK AUSTRIA 2014), we expect 50% of consumers likely to use the product rating. If 50% of these consumers show affection, we expect 25% of citizens will use the product rating system.
2. Retailers. They can pioneer future consumption patterns by offering products that are ethically more aligned to consumers’ sustainability preferences. They can focus on alternative business models: selling more highly rated products that might come with slightly higher prices (or profit margins).
3. Producers. They can offer more attractive, higher quality products to retail markets. Producers can sell higher quality products at a higher price. This incentivizes the improvement of production practices to align with consumer preferences regarding environment, health or worker rights.
4. Policy-making and institutions. New opportunities arise for environmental organizations to educate and interact with the general public by publishing their own AI personalization templates for consumers, e.g., adopting the WWF or Greenpeace sustainability priorities.
5. Global level/value: Food systems influence 12/17 SDGs of the UN. Their greenhouse gas emissions and the increase of the global population influence all UN geographical regions. The ASSET AI-based solution is a blueprint that assists citizens to push for a bottom-up shift to more sustainable consumption and production.
Scaling of AI solution: Please detail what proof of concept or implementations can you show now in terms of its efficacy and how the solution can be scaled to provide a global impact ad how realistic that scaling is.
A proof-of-concept has been demonstrated in real-world field tests in two supermarkets in Estonia and Austria. These pilots demonstrate how the AI-based product rating system achieves the ultimate goal of this project: its causal relation to a shift to more sustainable products choices at both supermarkets. The TRL-7 prototype used in the field tests consists of: the smart phone app that implements the AI-based product rating system, the sustainability ontology and knowledge-base, the database system and the products localization system in the supermarkets. The system was developed by an EU consortium that consists of partners such as a research center on mechatronics, the academic institutions of ETH Zurich and University of Leeds, a consumer organization, two retailers, service providers and with the support of organizations such as Greenpeace and other consumer organizations.
There are several alternative pathways for scaling up the concept globally. Communities and organizations could link products in the market to retailers that sell them so that the AI-based product rating system can operate at every retail shop worldwide. This process can also be automated to a large extent, e.g. collecting data from online shops. We managed with minimal resources, project experts and the involvement of citizens to create a complex sustainability ontology that covers a large spectrum of sustainability goals for thousands of products. We foresee a tremendous potential to scale up and maintain further such ontology with the involvement of relevant organizations, industrial partners and communities at a global level. The distributed design approach and open-source implementation of the AI-based product rating makes easier its applicability in different contexts as well as its further development. It also makes it more trustworthy and accountable and as such more stakeholders are likely to support it across the globe.
Ethical aspect: Please detail the way the solution addresses any of the main ethical aspects, including trustworthiness, bias, gender issues, etc.
The AI-based product rating system is by design privacy-preserving and distributed with its computations to limit use for manipulative nudging (marketing of certain products that do not justify sustainability). As a result, consumers can trust this AI solution and make use of it as public good given its open-source implementation. The explainability of the product ratings makes the content-based recommender system more accountable and transparent, while increasing the consumers’ awareness about sustainability aspects. This accountability and transparency is also evident in the co-design participatory approach for the co-creation of the sustainability ontology that involved experts, sustainability organizations as well as citizens’ crowd-sourced data and methods.
Developed on value-sensitive design principles, the ASSET product rating promotes both sustainability values in product choices as well as sustainable means to make these sustainable choices in terms of ethical principles and values in the design of the supporting technology. The system is also lawful and robust as successfully demonstrated in two real-world supermarkets in two different countries in the context of an EU project. The pilots have received ethical approval by the ethical committee of ETH Zurich.
The ASSET product rating system is by design an inclusive solution. It can be applied to physical shopping visits or even online for consumers with disabilities. The large spectrum of sustainability preferences (1025 combinations of sustainability choices) create a vast space of diversity and personalization with which consumers can express themselves and their priorities.