Attributing US productivity gains to AI is still an error, FT's data journalist argues
jburnmurdoch · x · 2026-11-01
FT data journalist John Burn-Murdoch argues it is currently an error to attribute US productivity data to AI. Even in finance, consulting and professional services, taking on more clients or projects tends to be labour-intensive and relational, so worker-level AI efficiencies don't necessarily translate into topline gains. In this sense software resembles manufacturing more than other services.
More from AGI Musings
- LeCun's new paper: 109M-param text JEPA beats 72B LLMs at aligning with vision — heyshrutimishra · 2026-11-02
- Yuval Noah Harari: AI is hacking the code of civilization — luisdans · 2026-11-02
- Ben Bajarin's reading list for understanding tech cycles and markets — BenBajarin · 2026-11-02
- AI is multiplicative, not additive: most people are multiplying by zero — mariofilhoml · 2026-11-02
- An AI agent just needs a prepaid card and compute? Reality is far messier — pramodk73 · 2026-11-02
- "Academy should drop its high horse on AI" sparks debate over silent campus adoption — TuhinChakr · 2026-11-02