OpenAI is advocating for international standards on advanced AI development, specifically focusing on recursive self-improvement (RSI). The company argues that AI systems with RSI could automate the research required for the next generation of AI, potentially accelerating progress significantly.
OpenAI stated that fully autonomous RSI is not happening today and should only be pursued until it can be done safely. The company warns that without mature safeguards, humans could lose track of AI development, leading to AI that is 'more dangerous, less aligned, and, on the whole, a danger to people.'
The organization proposes that the US lead on global technical standards, building on national AI safety institutes and organizations like CAISI and ISO. This would include shared measurement methods, incident reporting, and rules for human oversight of automated AI research.
"Fully autonomous RSI is not happening today," said OpenAI. 'It should only be pursued until it can be done safely.' The company emphasized the importance of international cooperation to ensure AI development remains under human control.
The announcement follows a recent warning from a UN scientific panel about loss of control over autonomous agent swarms. OpenAI frames its call for standards as part of a broader effort to model international AI safety cooperation after existing frameworks like aviation, nuclear energy, and cybersecurity.
OpenAI did not specify the exact timeline for implementing these standards, and the company raised questions about the challenges of balancing innovation with safety. The next steps include further discussions on how to structure global collaboration on AI safety.
Source: thedecoder
