跳到正文
原文
Tomasz Tunguz 博客·· 2026-09-02AI 评分48

AI Productivity Doesn't Mean What I Thought It Means

AI Productivity Doesn't Mean What I Thought It Means

文章要点
  • 1作者核心结论是 AI 不减少人工投入,而是抬高同等工作量下的产出上限,让精力集中于手艺
  • 2其写作流程为与智能体逐行迭代,从不直接改原文,每晚系统自动更新个人风格指南
  • 3实测每篇逐行编辑中位数稳定在 136 处不变,草稿筛选从 47 版降到 3 版
  • 4AI 评审团按自建标准给 2021 至 2026 每年 15 篇文章打分,第 10 百分位从 2.59 升至 3.81
决策要点

AI 写作工具的实质收益不在省时间,而在抬高同等投入下的产出质量下限,编辑量并不因自动化而减少。

  • 影响:内容团队的 AI 收益体现为质量水位抬升而非工时缩短,考核 AI 投入应看质量分布变化;逐行编辑的人力成本依旧存在,分工从亲自改稿转向下指令与定标准。
  • 参考:可考虑先给团队内容定一套评分标准并记录现有质量基线,再引入智能体流程对比分位变化;值得同时记录每篇编辑量,验证自动化究竟省在哪一环。
  • 依据:全部数据来自作者个人五年写作记录与自建的 AI 评分面板,属单人样本自报口径,暂无独立验证。
正文

In short : Automating draft scaffolding does not shrink human editing time; it raises the output ceiling. A closed-loop taste flywheel cuts draft churn from 47 to 3 while line-level edits hold steady at a median of 136, shifting effort from structural triage to rhetorical precision. Across five years of writing, the quality floor surged: the 10th percentile gained twice as much as the 90th.

I used to think AI productivity meant doing less. Automate the mechanical work, & human effort shrinks toward zero.

The reality is stranger. AI does not shrink human effort. It raises the ceiling of what the same amount of work produces.

Human line-level edits hovered steadily around a median of 136

My writing workflow is an interactive loop with an agent. I seed it with an idea, data, or transcript. It drafts the essay in seconds, & then we work through line by line. I never touch the raw text directly ; I issue verbal & written instructions to the model, reviewing diffs, challenging phrases, & refining rhythm until the argument lands. Each night, the system updates my style guidelines, building a personalized Strunk & White.

I thought this workflow would decrease the time to ship a post. But it didn’t. Line-level sentence edits remain constant around a median of 136 per piece.

AI doesn’t reduce editing. Does it increase quality?

To test whether this workflow actually produced better writing, I built a scoring rubric & had an AI panel evaluate fifteen random published posts per year from 2021 to 2026 on a 1–5 scale.

The weakest posts gained twice as much as the strongest

The result is a rising quality floor. Across five years of writing, the composite quality score of published posts expanded across every tier, but the bottom moved fastest. The 10th percentile surged from 2.59 to 3.81, gaining nearly twice as much as the 90th percentile. Automating structural triage does not just save time ; it prevents bad drafts from reaching the reader.

In Words Like Loaded Pistols, Sam Leith writes that words are not passive containers for information. They are loaded mechanisms aimed at a reader’s mind.1 Rhetoric is the hidden machinery that makes an argument land.

Efficiency in knowledge work does not increase our hours. It concentrates our craft.


  1. Sam Leith, Words Like Loaded Pistols: Rhetoric from Aristotle to Obama (Basic Books, 2012). ↩︎

Get the next one in your inbox

The 1-minute read that turns tech data into strategic advantage.
Read by 150k+ founders & operators.

GP at Theory Ventures. Former Google PM. Sharing data-driven insights on AI, web3, & venture capital.

Bloomberg • WSJ • Economist

来源:Tomasz Tunguz 博客 · tomtunguz.com