New MIT Research Shows Spectacular Increase In White Collar Productivity From ChatGPT
productivitychatgptmit-studywhite-collargenerative-ai
Abstraction: MIT RCT finds ChatGPT makes white-collar workers 37% faster at writing tasks
Key points:
- MIT study (not yet peer reviewed at time of writing) split 444 white-collar workers into ChatGPT and control groups for marketing, grant writing, data analysis, and HR writing tasks; work was graded by domain professionals
- ChatGPT group completed tasks 37% faster (17 minutes vs. 27 minutes) with similar quality; with iteration, quality improved significantly faster for the ChatGPT group
- ChatGPT primarily reduced rough draft creation effort; its role intensified slightly during final editing — confirming it dramatically accelerates "first draft" and initial findings phases
- Effect is equalizing: willingness-to-pay and quality gains were nearly identical for self-reported "bad writers" and "good writers," meaning the tool raises the floor and accelerates the ceiling
- Workers reported willingness to pay ~0.5% of annual salary for access — approximately $500/month for a $100K earner
- 68% of ChatGPT users submitted results from a single query, indicating people are not iterating heavily yet still seeing large gains
Connections: Chatgpt · Mit · Openai · Microsoft · AI Productivity · Generative AI