2022-03-28-世界经济论坛-Artificial_Intelligence_for_Children_38页_1mb

Summary of "Artificial Intelligence for Children" Toolkit

Core Content

This document, published by the World Economic Forum in March 2022, is a comprehensive toolkit aimed at guiding companies, product teams, parents, and guardians in developing and using AI technologies that are trustworthy, safe, ethical, and beneficial for children and youth. It emphasizes the need for child-centric design and outlines actionable steps to ensure AI systems respect the rights and development of young users.

Main Viewpoints

• AI is increasingly integrated into children's daily lives, from social media and education technology to smart toys and speakers.
• Children and youth are vulnerable to AI risks, including bias, privacy breaches, cybersecurity threats, and lack of accessibility.
• Responsible AI must be fair, inclusive, safe, transparent, and ethical, and should be designed with the child's best interests in mind.
• Companies have a responsibility to ensure AI products are suitable for children, considering their developmental stages and learning needs.
• Parents and guardians play a crucial role in selecting and supervising the use of AI products for children.
• The AI labelling system is proposed as a tool to inform consumers about how AI works, its potential impact, and the options available.

Key Information

1. C-Suite and Corporate Decision-Makers' Checklist

• Objective: Help companies design innovative and responsible AI for children and youth.
• Key Actions:
  • Know legal duties and regulatory constraints: Use existing guidelines (e.g., IEEE, UNICEF, WEF) and commit to internal and external AI oversight.
  • Build a diverse and capable team: Include ethicists, privacy specialists, educators, child development experts, and UX designers.
  • Train your team: Provide resources and education on responsible AI.
  • Contribute to AI regulations: Share insights and experiences to shape future standards.
  • Endorse AI labelling: Support and implement the AI labelling system to increase transparency and trust.

2. Product Team Guidelines

• Responsible design is an ongoing process throughout the product life cycle.
• Five core AI characteristics are emphasized: Fair, Inclusive, Responsible, Safe, and Transparent.
• Each characteristic includes:
  • Goals: What the AI should achieve.
  • Greatest potential for harm: Risks associated with the AI.
  • Risk mitigation: Strategies to address these risks.

Fair

• Ensure fairness and respect for user dignity.
• Address bias in training, expression, and feedback.
• Understand liability and how AI can be weaponized for harm.

Inclusive

• Design for accessibility and neurodiversity.
• Incorporate feedback from children and youth.
• Avoid exclusionary practices in AI systems.

Responsible

• Design AI that is age and developmentally appropriate.
• Reflect the latest learning science.
• Center children and youth in the design and development process.
• Overcommunicate privacy and security implications.

Safe

• Ensure the AI does no harm and cannot be misused.
• Prioritize cybersecurity and data protection.
• Mitigate risks of overuse and addiction.
• Build a transparent and user-driven relationship model between the child, guardian, and technology.

Transparent

• Explain AI functionality in non-technical terms.
• Clearly disclose use of high-risk technologies (e.g., facial recognition).
• Specify the age group the product is designed for.
• Provide guidelines for the environment in which the technology is used.
• Create alert mechanisms for guardians to intervene in case of risk.

3. AI Labelling System

• A physical and digital labelling system is proposed for AI products.
• Includes QR codes on packaging for easy access.
• Designed to provide clear, concise information about:
  • Age of the target users.
  • Data use and privacy practices.
  • Sensors and data collection methods.
  • Accessibility features.
  • AI use and its implications.
  • Networks and connectivity aspects.
• Aimed at helping consumers make informed choices and building trust with users and guardians.

4. Guide for Parents and Guardians

• Parents and guardians are key decision-makers in AI product adoption.
• Should be educated about AI benefits and risks.
• Use the AI labelling system to understand the AI's characteristics.
• Make deliberate and informed choices to protect children and ensure positive impact.

Conclusion

The toolkit aims to create a balanced and ethical approach to AI development for children and youth, ensuring that AI technologies are innovative, safe, and beneficial. It promotes a child-first mindset, urging companies to consider the psychosocial development, rights, and vulnerabilities of young users. By implementing the guidelines and labelling system, companies can lead in responsible AI, enhance brand trust, and contribute to a safer digital future for children.

References

• Benjamin, Ruha. Race after Technology: Abolitionist Tools for the New Jim Code. Polity Books, 2019.
• Coeckelbergh, Mark. AI Ethics. MIT Press, 2020.
• Dubber, Markus D., Frank Pasquale, and Sunit Das (eds). The Oxford Handbook of Ethics of AI. Oxford University Press, 2020.
• O'Neil, Cathy. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown Publishing Group, 2016.
• Russell, Stuart. Human Compatible: Artificial Intelligence and the Problem of Control. Penguin Publishing Group, 2019.
• United Nations. "Young people help draw up UN digital protection recommendations". UN News, 24 March 2021.