Supervision policies can shape long-term risk management in general-purpose AI models

2025-01-10

AI Summary

AI regulatory agencies struggle to maintain focus and comprehensively detect emerging community-level risks amid overwhelming volumes of incident reports. Method: This study develops a parametric simulation framework integrating millions of ChatGPT interaction logs with multi-source empirical data to systematically evaluate the long-term governance efficacy of four supervisory strategies. Contribution/Results: We first identify that supervision policies induce feedback loops in reporting behavior, systematically biasing risk perception. We propose a “diversity-prioritizing” strategy that dynamically balances high-risk response with broad-spectrum coverage. Experiments demonstrate that this strategy significantly improves mitigation of expert-identified high-impact risks but attenuates detection of community-reported systemic risks—empirically confirming that regulatory design directly shapes AI risk evolution pathways. Our work provides theoretically grounded, empirically validated frameworks and methodologies for building resilient AI governance systems.

Technology Category

• Philosophy and Ethics of AI: AI & Law, Justice, Regulation & Governance
• Multiagent Systems: Agent-Based Simulation and Emergent Behavior
• Humans and AI: Learning Human Values and Preferences

Application Category

• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

The rapid proliferation and deployment of General-Purpose AI (GPAI) models, including large language models (LLMs), present unprecedented challenges for AI supervisory entities. We hypothesize that these entities will need to navigate an emergent ecosystem of risk and incident reporting, likely to exceed their supervision capacity. To investigate this, we develop a simulation framework parameterized by features extracted from the diverse landscape of risk, incident, or hazard reporting ecosystems, including community-driven platforms, crowdsourcing initiatives, and expert assessments. We evaluate four supervision policies: non-prioritized (first-come, first-served), random selection, priority-based (addressing the highest-priority risks first), and diversity-prioritized (balancing high-priority risks with comprehensive coverage across risk types). Our results indicate that while priority-based and diversity-prioritized policies are more effective at mitigating high-impact risks, particularly those identified by experts, they may inadvertently neglect systemic issues reported by the broader community. This oversight can create feedback loops that amplify certain types of reporting while discouraging others, leading to a skewed perception of the overall risk landscape. We validate our simulation results with several real-world datasets, including one with over a million ChatGPT interactions, of which more than 150,000 conversations were identified as risky. This validation underscores the complex trade-offs inherent in AI risk supervision and highlights how the choice of risk management policies can shape the future landscape of AI risks across diverse GPAI models used in society.

Problem

• AI Governance
• Regulatory Oversight
• Risk Management

Innovation

• Regulatory Strategies
• AI Risk Management
• Bias Avoidance