世界经济论坛-重塑AI时代的法规：新西兰试点项目（英文）-2020.7-30页_1012kb

Summary of "Reimagining Regulation for the Age of AI: New Zealand Pilot Project" White Paper (June 2020)

Core Content

This white paper outlines a global initiative led by the World Economic Forum (WEF) in partnership with the Government of New Zealand to re-imagine AI regulation in the context of the Fourth Industrial Revolution. The project aims to co-design governance frameworks that are both effective and inclusive, ensuring that AI is used in a way that benefits society while mitigating risks.

Main Objectives

The project is structured around three focus areas:

1. National Conversation
2. Regulatory Capabilities and Institutional Design
3. Risk/Benefit Assessment of AI Systems for Government

These areas are designed to build trust, ensure ethical AI use, and provide a scalable regulatory approach.

Key Points

1. National Conversation

• Purpose: To obtain a social licence for AI use by engaging all stakeholders in an inclusive dialogue.
• Definition of Social Licence: The approval given by people to an organization they deem trustworthy enough to undertake potentially risky AI activities.
• Importance: Trust is essential for AI acceptance and must be built through transparency, inclusivity, and respect for human rights.
• Contextual Consideration: In New Zealand, the project must incorporate the Māori worldview and the principles of the Treaty of Waitangi.
• Recommendations:
  • Facilitate broad, collaborative conversations that respect diversity.
  • Frame national and global conversations on AI regulation in a coherent and accessible manner.
  • Use tangible and practical examples to show how regulation would work in practice.

2. Regulatory Capabilities and Institutional Design

• Objective: To develop in-house understanding of AI and produce well-informed regulatory policies.
• Approach: The project advocates for a multistakeholder, interdisciplinary approach to regulation.
• Regulatory Tools: A mix of hard and soft law mechanisms, including guidance, codes of practice, standards, certification, and co-regulation.
• Need for Agility: Traditional regulatory systems are insufficient for the rapid evolution of AI, requiring more agile and collaborative frameworks.
• Recommendations:
  • Develop regulatory capabilities through the design of an entity to oversee AI use.
  • Identify and iterate innovative approaches to assess algorithmic and AI use.
  • Conduct robust benefit and risk analyses.

3. Risk/Benefit Assessment of AI Systems for Government

• Purpose: To effectively mitigate risks associated with AI systems and maximize their benefits.
• Key Focus: Ensuring that AI systems are safe, reliable, and respectful of human rights throughout their lifecycle.
• Inclusive Design: The framework must be flexible enough for organizations and governments to adapt it to their specific needs.
• Recommendations:
  • Build a framework for risk/benefit assessment that is adaptable.
  • Prioritize accountability, fairness, safety, and accessibility in AI governance.
  • Use transparency as a key tool to build public confidence and encourage innovation.

Project Overview

• Multistakeholder Approach: The project involves collaboration between governments, companies, civil society, and academia.
• Global and Local Focus: While the project is based in New Zealand, it aims to create a framework that can be scaled globally.
• Three Focus Areas:
  • National conversation to gain social licence.
  • Development of regulatory capabilities.
  • Risk/benefit assessment of AI systems.

Project Timeline

• Scoping (September to December 2019): Building a core project community and identifying key issues.
• Co-designing (January to June 2020): Framing the conversation, identifying focus areas, and drafting governance frameworks.
• Testing (July to December 2020): Piloting new regulatory approaches and capturing lessons.
• Scaling (January 2021 onward): Encouraging broad adoption of the frameworks based on pilot outcomes.

Key Milestones

• Workshops:
  • New Zealand Workshop (October 2019): Focused on "How do we keep our citizens safe in the age of artificial intelligence?"
    • Key takeaways: Transparency, inclusivity, people-centred design, and human rights as core values.
  • Global Workshop (January 2020): Focused on "How do we achieve transparency and build trust with people in AI?"
    • Key takeaways: Trust through transparency, people-centred outcomes, and global collaboration.
• Webinars: Conducted during the pandemic to continue the project's work.
• Agreement: The three focus areas were refined and endorsed as the foundation for AI regulation.

Global Context

• OECD Guidelines: Adopted in 2019, these guidelines emphasize robust, safe, fair, and trustworthy AI systems and have been adopted by over 42 countries.
• United Arab Emirates: Released ethical principles for AI in 2019, emphasizing fairness, transparency, and accountability.
• European Union: The EC launched a white paper in 2020 and is developing a comprehensive regulatory approach, focusing on ethical and legal frameworks.
• United States: The Algorithmic Accountability Act was introduced in 2019, though it is still under consideration.
• Other Countries: Canada, Japan, Denmark, and Australia have also initiated AI-related policy work, often inspired by the OECD and EU approaches.

Need for Regulation

• Current Limitations: Existing regulatory systems are not fully equipped to handle AI's unique risks and benefits.
• Challenges:
  • Insufficient Regulation: Can lead to uncertainty, discrimination, and inequities.
  • Overregulation: May stifle innovation and create barriers to global interoperability.
• Ethical Considerations: AI regulation must be grounded in ethical principles, including fairness, transparency, and human oversight.

Conclusion

• The WEF and New Zealand aim to develop a regulatory framework that is inclusive, ethical, and adaptable.
• The project emphasizes the importance of transparency, social licence, and a people-centred approach to AI governance.
• The outcomes of this pilot will inform future global AI regulation efforts and encourage broader collaboration and engagement.