Why the science of AI can’t be left to scientists alone

The global AI governance challenge: The United Nations’ recent agreement to establish an “independent international Scientific Panel on AI” highlights the growing need for international coordination on artificial intelligence governance, but also raises questions about the most effective approach to achieve this goal.

• The lack of scientific consensus on AI risks and benefits has become a major obstacle for regulatory efforts at various levels of government, from local to international.
• World leaders are struggling to find common ground on AI policy, with differing perspectives emerging between countries like the United States and United Kingdom on one side, and France and China on the other.
• The need for a credible and globally legitimate scientific assessment of AI’s impacts is crucial for effective international coordination on AI governance.

Potential pitfalls of the UN’s approach: The emphasis on an “independent” scientific panel with minimal involvement from member states may lead to political irrelevance and limited impact on global AI policy.

• The current approach risks creating a panel that, while scientifically rigorous, may lack the political buy-in necessary to influence international decision-making on AI governance.
• This strategy contrasts sharply with more successful models of international scientific cooperation, such as the Intergovernmental Panel on Climate Change (IPCC).

Lessons from climate change governance: The IPCC’s model of involving governments throughout the scientific assessment process offers valuable insights for effective AI governance.

• The IPCC approach allows politicians to take ownership of scientific reports, leading to greater impact and relevance in policy-making circles.
• By involving governments in directing research, providing feedback, and endorsing findings, the IPCC has successfully bridged the gap between scientific expertise and political action.

A proposed alternative for AI governance: To achieve effective international coordination on AI, the UN panel should consider adopting a more inclusive approach that involves governments throughout the process.

• This strategy would involve member states in directing research priorities, providing feedback on drafts, and ultimately endorsing the panel’s findings.
• While this approach may result in slower, more politically negotiated outcomes, it is crucial for building global political buy-in for AI governance measures.
• The involvement of governments in the scientific assessment process can help ensure that the panel’s findings are relevant to and actionable by policymakers.

Balancing scientific integrity and political relevance: The challenge lies in striking the right balance between maintaining scientific rigor and ensuring political relevance in AI governance.

• Critics may argue that involving governments in the scientific process could compromise the independence and objectivity of the panel’s findings.
• However, political involvement is necessary to translate scientific knowledge into effective global policy action on AI.
• The IPCC model demonstrates that it is possible to maintain scientific integrity while also engaging with political realities and building consensus among diverse stakeholders.

Implications for the future of AI governance: The approach taken by the UN panel on AI could have far-reaching consequences for the development and regulation of artificial intelligence technologies globally.

• A purely scientist-led approach risks producing technically sound but politically irrelevant recommendations, potentially hindering effective international coordination on AI governance.
• Conversely, a more inclusive process that engages governments and other stakeholders could lead to more widely accepted and implementable AI policies across nations.
• The success or failure of this panel could set a precedent for how global challenges related to emerging technologies are addressed in the future.

Rethinking the role of science in policy-making: One must challenge the traditional notion that scientific assessments should be entirely independent of political processes, especially when dealing with complex global challenges like AI governance.

• In the case of AI, “science is too important to be left to the scientists” alone, and that political involvement is crucial for translating scientific insights into effective international coordination.
• This perspective calls for a reevaluation of how scientific expertise is integrated into policy-making processes, particularly for emerging technologies with far-reaching societal impacts.

Navigating the path forward: As the UN panel on AI takes shape, finding the right balance between scientific independence and political relevance will be crucial for its success and impact on global AI governance.