Scott Alexander: The Safe Uncertainty Fallacy in AI Predictions

moultano · x · 2026-08-22

Scott Alexander critiques the "Safe Uncertainty Fallacy"—the idea that "because a situation is completely unpredictable, it will be fine." He notes this logic was often used for AI timelines. Since 2017, AI has moved faster than expected, and GPT-4 arguably qualifies as AGI. When you have "no idea" when something will happen, sometimes the answer is "soon." Alexander反驳了 Tyler Cowen 等人用历史上印刷术或火的不可预测性来类比对 AI 的乐观态度。

Original post →

More from AGI Musings

AGI Musings channel →