Research map / Large language models / AI and society
Position: Technical Research and Talent is Needed for Effective AI Governance
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Position paper surveying EU, US, and Chinese AI policy documents to show gaps between regulatory technical requirements and current AI/ML research, and arguing for targeted technical research and more technical talent in AI governance.
Problem
Governments are enacting AI regulation that assumes technical tools (for data, compute, models, and outputs) that are underdeveloped or nonexistent, and governing institutions lack technical expertise to inform, operationalise, and enforce policy.
Approach
Maps specific technical requirements in EU AI Act, US E.O. 14110, AI Bill of Rights, and Chinese algorithmic recommendation and deep synthesis rules against the SOTA in AI/ML research to identify gaps, then argues for prioritised research and closer researcher-policymaker integration.
Key results
- Less than 1% of US AI PhD graduates choose government careers post-study, while over 65% enter the private sector and more than 25% pursue academia (cited data).
- The EU AI Act presumes systemic risk for GPAI models trained with more than 10^25 FLOPs, but no method exists to verify whether a large model is being trained.
- The paper identifies gaps across four technical categories (data, compute, models, output/deployment) and proposes two calls to action for the AI/ML community.
Surprising findings
- No method currently exists to verify whether a large model is being trained, making compute-threshold enforcement difficult.
- Less than 1% of AI PhD graduates in the US choose government careers after study.
Limitations
- Examples are illustrative rather than a comprehensive roadmap or prioritisation of issues, as the authors acknowledge.
- Covers only EU, US, and China policy documents, and omits the Chinese generative AI interim measures.
- Technical and talent solutions are only one component of governance; social science and public engagement fall outside scope.
Closest papers by meaning
- Position: If Innovation in AI systematically Violates Fundamental Rights, Is It Innovation at All? (NeurIPS 2025 · similarity 0.84)
- AI Governance and Lessons Learned as an AI Policy Advisor in the United States Senate (AAAI 2025 · similarity 0.83)
- Position: AI Safety should prioritize the Future of Work (ICML 2025 · similarity 0.82)
- Usage Governance Advisor: From Intent to AI Governance (AAAI 2025 · similarity 0.82)
- Position: The Right to AI (ICML 2025 · similarity 0.82)
- Resource Democratization: Is Compute the Binding Constraint on AI Research? (AAAI 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.