This research agenda is under construction and is not intended to be exhaustive.

AI-Driven Automation

Economic growth

• Endogenous economic growth model under advanced AI development

• In what ways can AI-driven automation lead to significant shifts in market dynamics, and how might this influence global competition and collaboration?

• Theoretical and practical circumstances of exponential growth rates due to AGI

• Compute-centric framework and take-off model à la Davidson report

• Assess the economic impacts of various AI risk scenarios. Develop risk metrics to assess the systemic impacts of AI, such as the risk of market manipulation in financial AI systems, or the risk of bias amplification in decision-making AI. 

• Applying graph theory to understand the diffusion of AI innovations across different critical industries and sectors

Labor market

• Income distribution and unemployment

Global Coordination

Compute Governance and Licensing Schemes

• Study of the potential incentive and regulatory challenges in implementing an international compute licensing system.

• How would a licensing scheme for computational resources impact market dynamics in the AI industry? Could a pricing model be employed and if so which one could be better?

• Game theory and strategic behavior in compute licensing: How might different AI stakeholders (companies, governments, research institutions etc.) behave strategically in response to a compute licensing scheme?

CERN for AI: Centralized vs. Decentralized AI Research Models

• Compare the safety, political and economic implications of centralized and decentralized models for AI research

• Evaluate the trade-offs in terms of risk management and AI progress

• Study of the impact of monopoly in frontier AI research on innovation in the broader market

Pause

• Investigate the immediate and cascading effects of a pause on frontier AI development

• How would a pause in AI development affect the global competitive dynamics of the AI industry? Analyze the strategic implications for countries and companies in the AI space, considering global leadership, investment shifts, and competitive advantages

• Study of scenarios of international conflicts in AI governance

Market Dynamics of Safe AI Development

• Investigate the role of monopolies, oligopolies, and competitive markets in advancing AI safety.

• Race dynamics and market structure: study of a "Race to the Bottom". 

• Antitrust Implications of AI: Exploring the role of antitrust laws in regulating AI markets.

• Infohazard policies and publications norms in AI capabilities

Ensuring Compliance With Standards

Self-Regulation and Certification:

• Investigate incentives driving companies towards self-regulation and certification in AI.

• Analyze the role of reputational costs, market signaling, and competitive advantage in promoting self-regulation.

• Auditing firms: study of the dynamics at play if more and more safety auditing firms in the market. Should we keep that market based or state based ?

Mandates and Enforcement by Supervisory Authorities:

• Use industrial organization theory to study how enforcement affects innovation incentives, market entry, competition and ultimately the level of risks across different failure mode scenarios.

Liability in AI Safety:

• Apply risk management and insurance economics to analyze how liability affects investment decisions and the development of safety features in AI.

• How can international liability norms be structured to govern cross-border AI incidents effectively?