Skip to content
Trending storyPast story

Dwarkesh Patel predicts math breakthroughs will automate AI R&D

1 report1 sourceupdated 3 days ago

What happened

AI digest

On the November 14 podcast, Dwarkesh Patel predicted that cascading progress in math will spill over into AI R&D, with AI quickly closing in on top human researchers. Citing math progress and verifiable metrics for pretraining, post-training and inference efficiency, he put the median year for automated AI R&D at 2028, with a 25% chance it slips past 2032. For the acceleration path after that, he gave 50% odds each to a self-accelerating software singularity and to a compute-limited short burst; in the latter case, 3–6 years of AI progress compresses into one year. These timelines and paths are his predictions, not achieved progress.

Written by AI from the coverage · updated 3 days ago

Coverage

Follow the reports to see the story from different sides.

Nov 14
  1. Dwarkesh Patel podcast
    The avalanche that hit math is coming for AI development

    Dwarkesh Patel 预测 AI 将迅速逼近顶尖人类研究员水平,并以 2028 作为实现 AI 研发自动化的中位年份判断。他以数学能力进展及预训练、后训练、推理效率的可验证指标支撑这一预测,同时给出 25% 的概率认为该节点会推迟至 2032 之后。他对持续自我加速的软件奇点与受算力限制的短期加速各赋予 50% 概率,后者预计将把 3–6 年的 AI 进展压缩到 1 年内。

Heat over time

Heat now 3·Comparable peak 10(Nov 14 04:00)·Comparable change over 24 hours -51%

02.557.510Nov 1404:00Nov 1419:00Nov 1511:00Nov 1602:00

The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.