Skip to content

R0007 — AI Made Everyone Faster

Fact-checking claims from "AI Made Everyone Faster. But Faster Is Not Necessarily Better." — covering individual performance distributions, toxic worker effects, AI leveling studies, and the gap between individual and organizational productivity gains.

Mode: Claim · Status: Active · Tags: ai, technology

Input

  1. O'Boyle and Aguinis (2012) studied five studies, 198 samples, 633,263 individuals across researchers, entertainers, politicians, and athletes and found individual performance follows a power-law distribution, not a normal distribution. The top decile produces roughly 30% of total output; the top quartile produces over 50%.
  2. O'Boyle and Aguinis won the Personnel Psychology Best Article award for this study.
  3. Their 2014 follow-up found that 82.5% of 229 samples had significantly heavy right tails.
  4. In software engineering, every major study from Sackman (1968) through Oliveira (2023) confirms large individual variation. The most careful recent work suggests log-normal distributions with roughly a 2.4x ratio between top and bottom halves.
  5. Schulmeyer formalized the "Net Negative Producing Programmer" concept in 1992 — programmers whose defect rates are high enough that the cost of their errors exceeds the value of their output. In a typical team of ten, he estimated up to three may qualify.
  6. Felps, Mitchell, and Byington demonstrated experimentally that a single negative team member reduces team performance by 30-40%.
  7. Housman and Minor studied 50,000 workers and found that avoiding one toxic hire saves $12,489 while hiring a top-one-percent superstar adds only $5,303.
  8. Toxic workers often have above-average raw output (Housman and Minor).
  9. No major enterprise survey (McKinsey n=1,933; BCG n=10,600; Deloitte n=3,235) identified capability-based stratification in AI deployment.
  10. Brynjolfsson, Li, and Raymond studied 5,172 customer service agents and found that low-skilled workers improved by 34% with AI, while experienced workers saw minimal gains.
  11. Noy and Zhang found the same leveling pattern in professional writing.
  12. Dell'Acqua and Mollick's study of 758 BCG consultants found that bottom-half performers improved by 43% versus 17% for the top half on tasks inside AI's capability frontier.
  13. In the same BCG study, consultants using AI on tasks beyond its capability were 19 percentage points less likely to get correct answers than those working without AI.
  14. The Otis study of Kenyan entrepreneurs gave GPT-4 business advice via WhatsApp. High performers gained roughly 15%. Low performers declined by roughly 8%.
  15. The DORA 2025 report found individual developers using AI completed 21% more tasks and merged 98% more pull requests, but organizational delivery metrics stayed flat.

Articles:

A0007 — AI Made Everyone Faster. But Faster Is Not Necessarily Better.

Published ~2026-03-11 · technology, ai

Runs

2026-03-20-02 — Second claim verification run

Mode: Claim · Claims: 15 · Prompt: claim v1.0-draft · Model: Claude Opus 4.6 (1M context)

15 claims re-investigated. 9 almost certain, 2 very likely, 4 likely. Four claims flagged for correction: C003 year attribution (2016 not 2014), C004 author attribution (Jorgensen not Oliveira), C005 possible earlier publication date, C009 McKinsey sample size mismatch. Strong convergence across AI leveling studies (C010-C014). DORA 2025 AI productivity paradox confirmed (C015).

2026-03-19 — Initial claim verification run

Mode: Claim · Claims: 15 · Prompt: Unified Research Standard v1.0-draft · Model: Claude Opus 4.6 (1M context)

15 claims investigated spanning performance distributions (C001-C004), toxic worker effects (C005-C008), enterprise AI deployment (C009), AI leveling studies (C010-C014), and organizational productivity paradox (C015). 8 claims almost certain, 2 very likely, 3 likely, 1 unlikely. Three claims flagged for correction (C003 year attribution, C009 stratification framing, C015 source attribution).