Anthropic robot study: 34% of US work hours feasible but only 0.3% cost-competitive
Crescitaly · reddit · 2026-10-01
Anthropic's September 30 study uses Claude to assess robot-executable tasks, operating environments and deployment costs. The poster urges scrutiny of two distinct numbers:
- Existing robots can perform tasks representing 34% of US working time in at least some settings
- But only 0.3% of working time is cost-competitive with human labor today
These are different measures, not forecasts of job loss; capability demonstrated in controlled facilities still counts.
The key gap to examine is between a rated task and a whole workflow: supervision, failures, handoffs, and tasks left to humans. The study authors warn that summing individual task costs can double-count robots or miss coordination costs.
The poster suggests checking sensitivity to per-task time, utilization, failure recovery, and human supervision before treating cost estimates as deployment decisions.
More from AGI Musings
- Mathematicians clash over whether AI labs should direct digestion of model proofs — burny_tech · 2026-10-01
- When anyone can code, the alpha shifts to problem identification, says Remi Louf — remilouf · 2026-10-01
- Founder: capital beats ideas in the AI age — agents need armies, secrecy wins — remilouf · 2026-10-01
- $2M manufacturing AI startup is just a dashboard with 3 OpenAI API calls — RexDouglass · 2026-10-01
- Greg Brockman says he no longer writes code himself as AI handles execution — Olivier__OG · 2026-10-01
- Millions of Consumer Agents: Why Aren't We Worried About Society-Level Agent Risk? — Natural-Pepper-2098 · 2026-10-01