IGF 2024 DC-CIV & DC-NN From Internet Openness to AI Openness

This session will focus on what lessons can be learned from Core Internet Values, particularly Internet openness, to foster AI openness.

Panellists will reflect on the past decade of Core Internet Values challenges to analyse technical, governance, and regulatory dimensions, with a particular focus on Internet openness, to understand what elements can be applied to the emerging field of AI governance to promote open AI and open AI governance.

Stakeholders will analyse a variety of experiences, including both policy and implementation, to understand what successes and failures drawn from Internet openness may inform AI openness.

Agenda

Introduction and Review of Core Internet Values (6 minutes)

• For newcomers who do not know what Core Internet Values are.

Presentation of Session Topics (6 minutes)

• Overview of the session’s focus on the parallels between Internet openness and AI openness.

Panellist Interventions (48 minutes)

• Panellists will reflect on the past decade of Internet openness challenges to analyse technical, governance, and regulatory dimensions.
• Discussion on how AI impacts core values such as best-effort, interoperability, openness, robustness, end-to-end, scalability, and permissionless innovation.
• Exploration of frameworks that can promote the empowerment of AI users, similar to net neutrality frameworks for Internet users.

Full Discussion with Participants (20 minutes)

• Stakeholders will analyse a variety of experiences, including both policy and implementation, to understand what successes and failures drawn from Internet openness may inform AI openness.
• Discussion on the specific issues raised by generative AIs in terms of Internet openness and how these new services affect end-user access to content online.

Conclusions (10 minutes)

• Summarising key takeaways and actionable insights.

Topics and Questions for Discussion

1. Lessons from Internet Openness for AI Governance

• What frameworks, if any, can promote the empowerment of AI users?
• To what extent can Internet infrastructure be compared to AI infrastructure?
• Can open Internet principles be translated to foster open AI? If so, what principles should be promoted?
• What have been the limits of net neutrality and Internet openness enforcement over the past decade, and how might these limits foresee upcoming bottlenecks in the AI value chain?

2. Core Values and AI

• How does AI impinge on core values such as best-effort, interoperability, openness, robustness, end-to-end, scalability, and permissionless innovation?
• How much does the healthy development of AI and its applications depend on these values?
• Can AI follow an open-source model?

3. Regulating AI and Internet

• What are the parallels in “regulating AI” with “regulating the Internet”?
• Can lessons learned from the defence of the Internet Model, with multi-stakeholder governance of resources, be applied to AI?

4. Proposals for Governance

• Bring all stakeholders to the table, including the technical community.
• Focus on concrete problems with the right combination of stakeholders and their specific representatives.
• Design institutional frameworks with representation, participation, transparency, accountability, and opportunities for reversal of decisions and redress.
• Ensure the ability to evolve by design and consider human agency in shaping conduct and impact.
• Look for a federated approach instead of a single top-down rule setter.