Three Ideas for Regulating Generative AI
normaltech.ai · 1,124 words · saved by 1 readers
Policy input to the federal government from a Stanford-Princeton team
Last week, we submitted comments to the National Telecommunication and Infrastructure Authority (NTIA), written in collaboration with Rishi Bommasani, Daniel Zhang, and Percy Liang at Stanford. In our comments, we describe the need for transparency around the generative AI ecosystem, the need for holistic public evaluations, and the guardrails needed for responsible open-source AI research and development. This post is a summary of our comments. Transparency about the digital supply chain that powers generative AI products Generative AI models are the building blocks for various digital…
related reading
- Radical Optionality — Governing Transformative AI Under Uncertaintyradical-optionality.ai
- Building Technology to Drive AI Governancebounded-regret.ghost.io
- How to pace the US frontierblog.aifutures.org
- Frontier AI Regulation | GovAIgovernance.ai
- [2304.04914] Regulatory Markets: The Future of AI Governancearxiv.org
- The AI rules that US policymakers are considering, explainedvox.com
- Executive Summary - AI Now Instituteainowinstitute.org
- What Should Be Donehyperdimensional.co
- On the Opportunities and Risks of Foundation Modelsarxiv.org
- Soft Nationalization: how the USG will control AI labs — LessWronglesswrong.com
- How Transparent Are Foundation Model Developers?normaltech.ai
- On the Societal Impact of Open Foundation Modelsnormaltech.ai