More Related Content


More from Steve Vosloo(20)


Why we need child-centred AI and how we can achieve it

  1. Why we need child-centred AI and how we can achieve it Steven Vosloo, UNICEF, Presentation at Beijing AI Conference, 1 June 2021
  2. KEY MESSAGES We need AI policies and systems to be child-centered We have ways to do this We all need to get involved
  3. The child rights framework Child rights in the digital era The CRC sets out the civil, political, economic, social, and cultural rights of children. The following rights are relevant to children in the digital world and set the framework for UNICEF to ensure the right of a child are considered in all aspects of digital work. CHILD RIGHTS: Acting in the best interests of the child The right to be protected from all types of exploitation & discrimination The right to obtain appropriate information The right to privacy and express their views The right to education, healthcare, etc. The right to leisure, play and culture The right to freedom of thought and expression CHILD RIGHTS
  4. UNICEF OFFICE OF GLOBAL INSIGHT AND POLICY AI and children: Opportunities • Aid children’s education and development • Contribute to better health outcomes for children • Support the achievement of the SDGs CHILD RIGHTS Smart toy Panda Bear by Fisher Price
  5. CHILDREN AND AI POLICIES AI and children: Risks and concerns • Systemic and automated discrimination and exclusion through bias • Limitations of children’s opportunities and development from AI-based predictive analytics and profiling • Infringement on data protection and privacy rights • Exacerbating the digital divide
  6. UNICEF OFFICE OF GLOBAL INSIGHT AND POLICY Opportunities or risks for how children’s rights are impacted by AI systems? • AI-enabled toys • Future of work • AI-powered voice assistants and chatbots • Facial recognition / analysis for biometrics CHILD RIGHTS Smart toy Panda Bear by Fisher Price
  7. CHILDREN AND AI POLICIES Why we need to focus on children • First generation that will never remember a time before smartphones. 1/3 of all online users are children • AI is in their lives – directly and indirectly • Children are not a homogenous group – they have developmental stages (early childhood, mid childhood, younger adolescence and older adolescence) • They are vulnerable and they have potential • → Affects their present and future
  8. 1.6.2021 AI FOR CHILDREN 8 Excerpts from child consultations “I am concerned about exactly the same thing that excites me, the fact that it accesses information, not just about things or places, but about people and that is where the fine line on people's privacy begins to break.” “It is also very important to educate from an early age what AI is, how it can help us and the care we should take.” “I am enthusiastic about the idea of being able to contribute in the future from my point of view as a woman and to be able to study a career in this area in order to develop inclusive artificial intelligence.” “Most of the technologies that exist are not made with children in mind.”.
  9. CHILDREN AND AI POLICIES Are current policies and practices enough? • If we just keep focusing on “human rights” or “vulnerable populations” is that sufficient to ensure children are considered in the policymaking process or practice of AI? • Do national AI policies, strategies and systems include children? • Are the 160+ AI principles enough to leverage the opportunities and mitigate the risks for children?
  10. CHILDREN AND AI POLICIES The AI for Children project • Partnership with the Government of Finland • How can AI systems uphold children’s rights? • Provide guidance to policy makers and corporations on how to create child-centred AI • Working together with the IEEE Software Association, Berkman Klein Center at Harvard University, 5Rights Foundation, World Economic Forum and others We do not have all the answers: • We put out the guidance for public consultation (summary of responses) • Invited anyone to pilot the guidance • Selected a few piloting organizations to learn from and write case studies on
  11. CHILDREN AND AI POLICIES Broad consultation and inputs • 5 expert consultation workshops with >200 participants in North America, Europe, Africa, Latin America and the Caribbean, and East-Asia and the Pacific • 9 child consultation workshops with 245 children consulted in Brazil, South Africa, Sweden & USA • Expert survey (33 responses, including from Cameroon, Jamaica, Nepal, Tunisia) + building on existing work, such as GDPR, RD4C, etc.
  12. New York: What excites me about AI?
  13. New York: What worries me about AI?
  14. South Africa: What worries me about AI?
  15. Child-centred AI Support children's development and well-being Ensure inclusion of and for children Prioritize fairness and non-discrimination for children Protect children's data and privacy Ensure safety for children Provide transparency, explainability, and accountability for children Empower government and businesses with knowledge of AI and children’s rights Prepare children for present and future developments in AI Create an enabling environment for child-centred AI Uphold children’s rights Through the lenses of protection, provision and participation Requirements Foundation
  16. CHILDREN AND AI POLICIES Support children’s development and well- being Let AI help me develop to my full potential. • Design for a child rights approach • Improved child well-being as a primary success criterion for system quality • Environmentally sustainable AI • Pilot: Milli chatbot by Helsinki University Hospital (Finland) • AutismVR by Imìsí 3D (Nigeria)
  17. CHILDREN AND AI POLICIES Ensure inclusion of and for children Include me and those around me. • Meaningful child participation, both in AI policies and in the design and development processes • Diversity amongst those who design, develop, collect and process data, implement, research, regulate and oversee AI systems
  18. CHILDREN AND AI POLICIES Prioritize fairness and non-discrimination for children AI must be for all children. • Support the most marginalized children, including girls, children from minority or marginalized groups, children with disabilities and those in refugee contexts • Develop datasets so that a diversity of children’s data are included • Pilot: Alerta Niñez by Ministry of Social Development (Chile)
  19. CHILDREN AND AI POLICIES Protect children’s data and privacy Ensure my privacy in an AI world. • Responsible handling of children’s data • Adopt a privacy-by-design approach • Special protections for marginalized groups and for particularly sensitive data, including ethnicity and biometric data
  20. CHILDREN AND AI POLICIES Ensure safety for children I need to be safe in the AI world. • Initial and ongoing child-rights impact assessments • Leverage the use of AI systems to promote children’s safety • Pilot: SomeBuddy (Finland and Sweden)
  21. CHILDREN AND AI POLICIES Provide transparency, explainability, and accountability for children I need to know how AI impacts me. You need to be accountable for that. • Use age-appropriate language to describe AI • Keep a human in the loop for key life decisions that impact children, medical diagnoses, welfare decisions or processing school applications • Review, update and develop AI-related regulatory frameworks to integrate child rights • Establish AI oversight bodies for child rights and set up support mechanisms for redress • Pilot: Haru robot by Honda Research Institute Japan & Joint Research Centre of the European Commission (Japan and Uganda)
  22. CHILDREN AND AI POLICIES Empower governments and businesses with knowledge of AI and children’s rights You must know what my rights are and uphold them. • Commit to child-centred AI and put in place mechanisms to realize this in practice • Pilot: Responsible AI framework and assessment checklist / H&M Group (Sweden, global) • Capitalize on customers’ demand for trusted and transparent AI solutions for children
  23. CHILDREN AND AI POLICIES Prepare children for present and future developments in AI If I am well prepared now, I can contribute to responsible AI for the future. • Teaching technical and soft skills needed to flourish in an AI world, including in the future workplace • Leverage the use of AI systems in education, when it is appropriate • Develop and promote awareness campaigns for parents, caregivers and society as a whole.
  24. CHILDREN AND AI POLICIES Create an enabling environment Make it possible for all to contribute to child- centred AI. • Infrastructure and digital ecosystem investment to limit the “AI divide” • Provide funding and incentives for child-centred AI policies and strategies • Support research on AI for and with children • Digital co-operation • Pilot: AI Sweden (Sweden)
  25. CHILDREN AND AI POLICIES EC Proposal for a Regulation of AI • Recognizes the rights of every child (CRC) • AI should not exploit vulnerabilities of children and people due to their age, physical or mental incapacities • Assessment of AI’s impact on children • List of prohibited AI: The prohibitions covers practices that have a significant potential to manipulate persons through subliminal techniques beyond their consciousness or exploit vulnerabilities of specific vulnerable groups such as children or persons with disabilities in order to materially distort their behaviour in a manner that is likely to cause them or another person psychological or physical harm.
  26. CHILDREN AND AI POLICIES Recognising the specific challenges and opportunities AI presents for children, we will adopt UNICEF’s policy guidance on AI for children which draws on the UNCRC
  27. CHILDREN AND AI POLICIES Why we need AI policies and systems to be child-centered We have ways to do this We all need to get involved
  28. CHILDREN AND AI POLICIES Connect with us Global Forum on AI for Children, Nov 2021 Subscribe to project mailing list: Email: Web: