Open Letter to Amodei and Altman Challenges Frontier AI Slowdown Calls: State Keeps Moving
AryHHAry · x · 2026-09-14
- 作者在致 Dario Amodei、Sam Altman 等前沿实验室 CEO 的公开信中承认「前沿应当放缓、安全伦理应先行」的立场,但反驳书面呼吁本身无约束力。
- 核心论点:在信息物理学意义上,增加一个观察通道(带徽章的评估员)不等于掌握系统状态;模型权重、芯片、下一次训练、MIT 许可的开源 checkpoint 才是状态,而「速度限制只约束那些可以被起诉的人」。
- 作者提出具体诉求:赋予嵌入式评估员发布权、在能力下一次跃升之前资助对齐工作、把伦理作为训练运行的约束而非发布时附加的段落,并要求所有参与竞赛的实验室都如此,包括不写公开文章的那些。
Related event: Dario's Call to Slow Frontier AI Sparks Global Governance Debate(39 posts)→
More from AGI Musings
- Economists warn: AI agent-based modeling risks repeating the failed SFI agenda — Afinetheorem · 2026-11-01
- "Funny that no one knows how to make money from intelligence" — gklambauer · 2026-11-01
- Agentic AI is training users to articulate their desires and principles precisely — sebpaquet · 2026-11-01
- Gartner: 38% of orgs pilot AI agents, only 11% in production, 40% of projects to fail by 2027 — ingliguori · 2026-11-01
- Weekly: Anthropic's S1 filing, labs claim models refute Exponential Time Hypothesis — njyx · 2026-11-01
- Pedro Domingos revives Moravec's paradox: AI cracks Navier-Stokes but can't do accounting — pmddomingos · 2026-11-01