AI leaders warn self-improving models could trigger an “intelligence explosion”
A new white paper backed by figures including Geoffrey Hinton and senior leaders at OpenAI, Microsoft and Anthropic reportedly warns that AI systems automating their own improvement could accelerate progress so fast that governments have little time to react.
TLDR
A new white paper, as reported by Axios, warns that AI models capable of improving themselves could set off an “intelligence explosion,” sharply accelerating AI progress. Axios reports the authors include Geoffrey Hinton, OpenAI’s chief scientist, Microsoft’s chief scientific officer and an Anthropic co-founder. The paper points to signs that systems at Anthropic and OpenAI are already taking on more internal R&D and engineering work, and estimates that expert-level AI R&D could let one developer direct an AI workforce comparable to millions of top human researchers. The authors reportedly say the outcome is not certain, but warn that if it begins, the chance to respond may close quickly.
Combined views
16.3K
2 Sources, first seen 4h ago