The Last Ten Per Cent
Summary
The article analyzes the economic implications of artificial intelligence automating "90 percent of a job," a concept introduced by Anthropic CEO Dario Amodei in May 2026. It explores whether the remaining "last ten percent" of work expands into more productive roles or merely reduces job scope, potentially decreasing employment. The analysis advocates for a task-level approach, differentiating between technology's displacement of existing tasks and its reinstatement of new human-comparative advantage tasks. Key considerations include how automation affects expertise thresholds, potentially making jobs more demanding or accessible, and the "focus mechanism," where freed time enhances the quality of remaining human-performed tasks. The piece also examines how firms redesign job roles post-automation and notes that traditional payroll data often fails to capture shifts in work intensity or quality. It concludes by proposing six critical questions for decision-makers to assess AI's true impact beyond initial exposure estimates.
Key takeaway
For business leaders evaluating AI integration, you must look beyond simple task automation percentages. Instead, analyze how remaining tasks evolve, considering shifts in required expertise and potential for human focus to improve quality. Actively redesign job roles to optimize for new task bundles or specialization. Crucially, assess demand elasticity and how automation impacts work intensity, as payroll records alone won't capture the full economic or human impact.
Key insights
AI automation's impact on jobs depends on how remaining tasks evolve, not just task displacement.
Principles
- Technology changes specific activities, not whole job titles.
- Automation involves displacement and reinstatement effects.
- Firms actively redraw job roles post-automation.
In practice
- Assess tasks as substitutes, complements, or bottlenecks.
- Analyze how automation changes expertise thresholds.
- Evaluate demand elasticity for services post-automation.
Topics
- AI Automation
- Labor Economics
- Task-Based Analysis
- Job Redesign
- Workforce Transformation
- Expertise Thresholds
Best for: Research Scientist, Policy Maker, Executive
Related on AIssential
See Counsel's argued verdicts on the open AI decisions leaders are weighing →
Editorial summary, takeaway, and curation by AIssential. Original article published by Agglomerations.