本文由 AI 分析生成
建立時間： 2026-09-06 來源： https://x.com/Konstantine/status/2095166660223397963
Summary
The essay argues that AI represents a “Cognitive Revolution” that rhymes with the Industrial Revolution’s mechanization of physical work: just as physical work went from ~99% biological to ~99.9% machine over two centuries, cognitive work is on the same trajectory but compressed into decades. It walks through the parallels — input scaling (coal/steel → electricity/compute), demand explosion via price collapse (Jevons’ paradox), the “primal application” (textiles → AI coding), employment disruption (Engels’ pause), and effects on education, health, and daily life — concluding that what remains distinctly human is “wanting things, choosing between them, being accountable for the choice.”
文章主張 AI 代表一場與工業革命（機械化取代體力勞動）相呼應的「認知革命」：正如體力勞動在兩百年間從約 99% 由生物完成轉變為 99.9% 由機器完成，認知工作正走上同樣的軌跡，卻被壓縮進短短幾十年。文章逐一鋪陳其對應關係——投入規模擴張（煤鐵 → 電力與運算）、需求因價格崩跌而爆發（傑文斯悖論）、「原始應用場景」（紡織 → AI 編程）、就業衝擊（恩格斯停頓期）、以及對教育、醫療與日常生活的影響——最後總結：真正留給人類的是「想要什麼、如何選擇、並為選擇負責」。
Key Points
- Core parallel: physical work went 99% biological → 99.9% machine over ~200 years (1700s–present); cognitive work is argued to follow the same curve but faster, since “intelligence per Watt” is falling >10x/year vs. mechanical work’s slower 19th-century price collapse
- AI coding is framed as this revolution’s “Spinning Jenny” (the primal, first-commercialized skilled application) — not because it’s the most important eventual use, but because it’s where the human moved from performing the task to supervising it first
- Jevons’ paradox applied to cognition: cheaper thinking won’t reduce total demand for it, it will multiply it — “when cognition costs approach zero, every curiosity can get a research team”
- Employment framing via “Engels’ pause”: 1780-1840 UK saw output/worker +46% but real wages only +12% (workers lagged productivity for a generation) before wages caught up and then outran productivity 1840-1900 — the essay argues AI’s version of this pause may be much shorter (a career, not a generation) which raises the stakes for retraining/safety nets
- Argues mass education and its “factory model” (bells, rows, fixed schedules) were themselves industrial-era artifacts training workers for synchronized factory shifts — cheap AI tutoring undermines the rationale for that model (cites Bloom’s 1984 finding that 1:1 mastery tutoring outperforms 98% of classroom instruction)
- What’s argued to remain human longest: things that were “never really cognition” — wanting, choosing, being accountable for a choice, being trusted by others (“the machine can draft the treaty, someone still has to sign it”)
Insights
- The Jevons’ paradox framing is a sharper and more falsifiable claim than typical “AI will/won’t take jobs” takes — it predicts more total cognitive labor demand as AI gets cheaper, which is testable against sector-level hiring data (the essay itself references, without fully presenting, inflection data in software/legal/financial hiring)
- The “Engels’ pause” analogy is doing a lot of work to acknowledge near-term pain (job displacement) while still landing on a long-run-optimistic conclusion — worth noting this is explicitly a historical-analogy argument, not an empirical projection, and the essay’s own caveat (“this revolution may offer five years, not forty”) undercuts the reassurance somewhat
- Connects to “Line Switching” and general AI-era productivity essays clipped in this batch — the recurring theme across several is that AI commoditizes shallow/discoverable work and raises the value of depth, judgment, and accountability specifically
Connections
Raw Excerpt
What stays economically human the longest is what was never really cognition to begin with: wanting things, choosing between them, being accountable for the choice, and being trusted by other people. The machine can draft the treaty. Someone still has to sign it.