AI Sweden and UNICEF: 

The Intersection of Children’s Rights and AI Innovation

Grace Weems

Introduction

For two years, the United Nations Children’s Fund (UNICEF) built a project to analyze the use, integration, and effects of Artificial Intelligence (AI) in children’s lives. The UNICEF website stresses the importance of understanding how AI affects children and how we can create “child-centered AI” that protects children as well as provides and empowers them (UNICEF, “AI for Children”). Per UNICEF’s policy guidance, public and private sectors must support development, ensure inclusion to promote fairness and non-discrimination, protect children’s data and privacy, ensure safety, provide transparency and accountability for children, and help prepare children for the current and future AI developments (Dignum et al., 2021). To promote this project, UNICEF worked with different businesses, corporations, and governments to ensure equality in child-centered AI. As one of the top leaders in digital transformation, Sweden recognizes the potential of AI in regard to social, economic, and environmental challenges. The Swedish government also recognizes issues with AI regarding transparency, data and privacy, and discrimination and wishes to lay a foundation for a national approach to AI (Ministry of Enterprise and Innovation, 4).

Artificial Intelligence has become a massive industry around the world as technology continues to advance and improve. AI has been integrated into business, research, education, and medical care. Furthermore, even governments are looking into AI in regard to its problems, solutions, advancements, and possibilities. In 2020, Sweden incorporated the United Nations (UN) Convention on the Rights of the Child (CRC) into law with the goal of implementing the CRC into policies. For the most part, municipalities in Sweden have been successful in incorporating the CRC, except AI has posed many problems for integration (Pauwels, 2). AI is somewhat new and ever-changing, so implementation and support are precarious.

National Approach

Sweden has a decent track record regarding privacy and data collection policies, laws, and regulations. In fact, in 1973 Sweden was the first country to pass a law protecting personal data on computers (Johnny 5, 2016). Furthermore, when the EU passed a directive that required EU states to collect and retain data from its “customers,” Sweden refused to implement the directive and was eventually taken to court by the European Commission; in the end, the Swedish government ended up implementing the directive in 2012 (Johnny 5, 2016). Furthermore, Sweden passed the Sweden Electronic Communications Act in 2003, which ensured users secure access to the internet (Johnny 5, 2016). With such a history of prioritizing the privacy of its people, it’s not shocking that privacy concerns over AI have become an important issue in Sweden.

As AI began growing and including itself more into industries, sectors, and everyday life, Sweden realized its need to not only promote AI, but set regulations and ensure that AI is being used fairly and ethically. Sweden prides itself on being a vanguard of digital transformation and believes it to be their responsibility to promote the use and development of AI. In the official National Approach to AI, the Swedish government stresses the need for AI application in education, research, innovation, and infrastructure (Ministry of Enterprise and Innovation, 5). They also stress the importance of data access, safety, and ethics in AI application (Ministry of Enterprise and Innovation, 10).

In regards to education, Sweden finds it important for higher education institutions to implement Ai learning in curricula so incoming professionals are able to properly use AI in their workplace (Ministry of Enterprise and Innovation, 6).. Additionally, the Swedish government wishes to promote both research in AI as well as AI in research. The national approach also encourages innovation in AI to be able to implement AI in a variety of different sectors, while also managing any risks AI might bear (Ministry of Enterprise and Innovation, 7). AI must be transparent, accessible, and follow rule of law procedures (Ministry of Enterprise and Innovation, 8). Lastly, Sweden’s approach to AI focuses on approaches, rules, and standards to the ever-changing digital framework and infrastructure to ensure safety, sustainability, and value. Much of this concern surrounds data and privacy, in which Sweden emphasizes the EU’s General Data Protection Regulation from 2018 that provides users with strong privacy protection (Ministry of Enterprise and Innovation, 10).

AI Sweden & Child-Centered AI

AI Sweden is Sweden’s national center for AI. They are funded by the government as well as public and private partners. Looking at their list of partners, AI Sweden has been able to attract some of the largest corporate names — Microsoft, Google, Dell, and Volvo (“Partners”). According to their website, AI Sweden’s mission is to accelerate and broaden AI’s usage in society. Some of their largest projects focus on climate change, language, science, and journalism (“About AI Sweden”). Another tool AI Sweden provides to private and public sectors is the Data Factory, which allows partners to make use of data as well as contribute their own (Pauwels, 5).

In 2020, UNICEF and AI Sweden committed to a joint project to analyze the CRC and UNICEF’s AI policy guidance in child-centered AI. The first step of this project was set up in three Swedish municipalities: Malmö, Lund, and Helsingborg. This project, known as the Three Cities Initiative, was used to study how stakeholders in these cities were able to collaborate and learn about children’s rights through child-centered AI (Pauwels, 3). After this study, the national track began with the Swedish Innovation Agency as the sponsor and the collaboration between AI Sweden, Mobile Heights, aiRikr Innovation, and Lund University. The national track’s goal was to create an initial framework for child-centered AI by examining components needed through analyzing past studies, specifically the Three Cities Initiative study (Pauwels, 3).

According to the CRC, which Sweden has implemented into law, “children shall be protected against all forms of discrimination, the best interest of the child shall be the primary concern in all actions concerning children” (Pauwels, 4). These policies have been valuable to Sweden in regards to AI for children; however, the quickly changing technological innovations have made it difficult to ensure equal inclusion of children in AI (Pauwels, 4). One of the main reasons it has been so difficult is due to the fact that technological innovation is under the Department of Infrastructure while children’s well-being is under the Ministry of Health and Social Affairs, so legislative change and progress can be slow and inconvenient (Pauwels, 5). AI Sweden believes that this study will help Sweden create a framework for child-centered AI to be more equal, equitable, fair, and safe.

AI Sweden focused on empowering businesses and governments with “knowledge of AI and children’s rights, and to create an enabling environment for child-centered AI” (Pauwels, 6). To achieve this, the study focused on four methods: “1) create awareness and deepen understanding, 2) educate, 3) promote networks and collaboration, and 4) design adequate policies and tools” (Pauwels, 6). Before any progress can actually be made, it’s important for institutions and organizations to understand child-centered AI. Therefore, AI Sweden stressed the importance of creating awareness and deepening understanding as the first step in this study (Pauwels, 6). Furthermore, AI Sweden found that educating people about AI development and children’s rights and the intersectionality between the two is key in implementing child-centered AI policies. Another method AI Sweden stressed in this study is a collaboration between networks. By involving all parties — private, public, government, and NGOs — Sweden will be able to achieve a more well-rounded approach to these policies (Pauwels, 7). Lastly, this case study’s goal is to be able to make a national framework for child-centered AI development (Pauwels, 8). Through these steps, methods, and measures, AI Sweden hopes to help create policies that suit UNICEF’s CRC guidelines in the AI sector.

The study focused on 3 major cities in Sweden: Helsingborg, Lund, and Malmö. As the study began to be implemented in these cities, AI Sweden discovered that although governments and NGOs were already planning and studying the connection between AI and children’s rights, they often faced obstacles to passing any effective policies. These obstacles are in-large part due to the inability of child development advocates and AI developers to collaborate and function together (Pauwels, 9). In order to create a well-rounded study, each city focused on a different aspect of child-centered AI development. In Helsingborg, the focus was on an AI chatbot, LAIban, used by preschoolers. The chatbot is accessible in the classroom via tablet and students can answer question prompts provided, play, and learn. It also offers several different languages. For this study, LAIban was examined on how well it adhered to three requirements. First, it needs to support children’s development and well-being. LAIbot “offers guidance and information on several levels — ranging from daily help to long-term learning” and information ranges from weather, lunches, lessons and games, and even sustainability. Second, LAIbot needed to ensure the inclusion of and for children and avoid any discrimination, which the study concluded that it does. Lastly, the tool needs to be able to provide transparency and accountability (Pauwels, 11). The city of Lund focused on the process of shifting child-centered AI requirements into policy based on the “Barnets Bästa” policy, which means “best for the child”. The Barnets Bästa was revised in order to follow UNICEF’s child-centered AI policy guidances. Such revisions include a renewed emphasis on child safety, inclusivity, security of data, and transparency (Pauwels, 13). The third city was Malmö, which studies the social impacts of and on AI in different socioeconomic areas of the city. As of now, they are working on finding the necessary resources to analyze the full impact (Pauwels, 13). The city did find that there is tension in the balance between protecting children’s rights of data protection and privacy and using AI to support children’s long-term well-being (Pauwels, 14). Through these three cities’ findings, AI Sweden and UNICEF were able to look at the bigger picture of how to implement child-centered AI development policies.

Works Cited

About AI Sweden. (2022, January 13). AI Sweden. https://www.ai.se/en/about-0.

Dignum, V., Penagos, M., Pigmans, K., & Vosloo, S. (2021, November 19). Policy guidance on AI for children. UNICEF Office of Global Insight & Policy. https://www.unicef.org/globalinsight/reports/policy-guidance-ai-children.

Johnny 5. (2021, October 4). Online privacy is why Sweden wins the internet. ExpressVPN. https://www.expressvpn.com/blog/online-privacy-is-why-sweden-wins-the-internet/.

Ministry of Enterprise and Innovation. (2019, February). National approach to artificial intelligence. Government Offices of Sweden. https://www.government.se/4a7451/contentassets/fe2ba005fb49433587574c513a837fac/national-approach-to-artificial-intelligence.pdf.

Partners. (2022, April 6). AI Sweden. https://www.ai.se/en/partners-9.

Pauwels, E. (2021, November). AI Sweden & three cities. UNICEF Office of Global Insight & Policy. https://www.unicef.org/globalinsight/media/2346/file.

UNICEF. AI for children. UNICEF Office of Global Insight & Policy. Retrieved March 28, 2022, https://www.unicef.org/globalinsight/featured-projects/ai-children#Pilot%20case%20studies.