ARTFEED — Contemporary Art Intelligence

Economists Argue AI Reshapes Work Through '90/10 Production Function'

ai-technology · 2026-10-02

An essay in The New Yorker explores the evolving dynamics of the workplace due to AI, referencing the research paper "Messy Jobs: The Work That AI Cannot Reach" by economists Luis Garicano, Jin Li, and Yanhui Wu. They introduce the concept of a "90/10 production function," suggesting that while AI can achieve 90% quality, it struggles to reach excellence. A study indicates that artists utilizing AI tools can produce work in 30 minutes that would typically require two hours manually, with further adjustments yielding diminishing returns. The piece differentiates between "commodity tasks" and "star tasks," analyzing AI's effects on junior employees. It notes BBVA's successful AI adoption and mentions a decline in software engineer job openings in 2025, followed by an increase in 2026. Researchers from Stanford cautioned against making broad conclusions from initial AI work-related evidence.

Key facts

  • Economists Luis Garicano, Jin Li, and Yanhui Wu authored "Messy Jobs: The Work That AI Cannot Reach"
  • The "90/10 production function" describes AI making it easy to reach ninety-per-cent quality but hard to achieve excellence
  • In a study, artists using AI reached in half an hour a quality level that would have taken two hours by hand
  • ChatGPT first appeared in 2022; Claude in 2023
  • Job openings for software engineers decreased substantially in 2025 and increased in 2026
  • A group of Stanford researchers cautioned in July that early evidence is hardly the last word on the future of work in an AI world
  • The Spanish bank BBVA has eighty million customers and more than a hundred and twenty thousand employees
  • The authors distinguish "weak bundles" from "strong bundles" of tasks, in which separation destroys value

Entities

Artists

  • Luis Garicano
  • Jin Li
  • Yanhui Wu
  • Claudette
  • Frank
  • Jill
  • Geoff
  • Linda

Institutions

  • Stanford University
  • BBVA
  • The New Yorker
  • OpenAI
  • Anthropic

Locations

  • Spain
  • United States

Sources