OpenAI CEO Sam Altman and Elon Musk have endorsed Dario Amodei's call for independent oversight in AI labs, aligning with Ex-Deepmind-CEO Demis Hassabis. All three leaders agree that AI development should be subject to external evaluation to ensure safety and ethical standards.
Altman also confirmed that OpenAI will not go public this year, citing ongoing safety concerns. He had previously shared this decision internally in June, suggesting a strategic pause to prioritize responsible development.
Google researcher Peyman Milanfar challenged Amodei's 'speed limit' proposal, arguing that recursive self-improvement (RSI) is an unstable assumption. He contended that reliable improvement requires real-world feedback, not just benchmarks, making external oversight less critical.
"Systems that reliably improve themselves will be governed, damped, slow, and bounded by margins that look wasteful. Because stability is the speed limit," said Peyman Milanfar, a Google researcher. Milanfar emphasized that self-improvement in AI systems must be tempered by real-world constraints.
The debate follows growing concerns about AI safety and the potential risks of rapid development. The source highlights the tension between innovation and regulation in the AI field.
OpenAI did not comment on the specifics of Amodei's oversight proposal, and the open question remains about how to balance speed with safety in AI development. The source notes that the discussion will continue as AI research progresses.
Source: thedecoder
