OpenAI Calls for US Leadership in Global AI Standards
OpenAI wants the United States to take the lead in establishing international standards for self-improving Artificial Intelligence (AI) (164news.com's guide to artificial intelligence (ai)), including systems that can develop their own successors. This proposal comes during a week filled with AI diplomacy, with Sam Altman briefing the UN Security Council on Wednesday and a meeting between Xi Jinping and the White House on Thursday.
"Fully autonomous recursive self-improvement (RSI) is not happening today, and we should not pursue it unless and until it can be done safely," OpenAI stated in their proposal published on September 21st.
Key Points:
- Standard Setting: OpenAI suggests building upon existing AI safety institutes worldwide, with the US Center for AI Standards and Innovation (CAISI) playing a central role. These standards would focus on frontier AI models and their benefits and risks, without mandating pre-release reviews or approvals.
- Three Focus Areas:
- Measuring RSI progress and automated research within AI companies.
- Establishing human oversight guidelines for specific automated research processes.
- Developing a common system for reporting and classifying AI incidents with defined severity levels. OpenAI offers its misalignment reporting framework as an initial contribution.
- International Collaboration: The company advocates for secure channels between governments and critical infrastructure operators to share threat intelligence, highlighting the potential for dialogue between the US and China.
Why the US?
"The United States is well positioned to lead because its AI industry is at the technical frontier, and it still holds a privileged global network position in areas such as finance, trade, defense, technology, and information systems," OpenAI explains. They believe leading now will determine whether the US shapes the global AI framework or observes a fragmented system emerge.
This proposal comes during a significant week for AI diplomacy, adding to ongoing discussions around AI's future.