The capability half-life: skill decay rates in an agentic economy
Empirical study of how quickly professional skills decay when AI systems can perform those tasks, and evidence-based strategies for continuous reskilling.
Authors
R. Nakamura, DRUCKER, 15 contributors
Published
2030
Citations
312
Overview
In an economy where AI agents can perform tasks at scale, professional skills that were valuable become obsolete faster. But which skills decay fastest? Which remain valuable? This research tracked skill decay across 14 professions over 2.5 years, measuring both direct skill relevance and skill-adjacent competencies.
Methodology
Cohort study tracking 2,400 professionals across 14 professions over 2.5 years. Quarterly skills assessments using validated competency frameworks. Labor market tracking through job postings, salary signals, and hiring patterns. Case studies of 60 professionals making successful career transitions.
Key Findings
Task-specific skills (e.g., specific coding language, specific financial analysis technique) have half-lives of 18-24 months when AI systems enter the market for that task. But meta-skills (learning ability, systems thinking, communication) show no measurable decay and may actually appreciate in value as domain knowledge becomes less scarce.
The highest-decay-risk professions are those where tasks are highly structured and well-documented (data analysis: 16-month half-life, bookkeeping: 14-month half-life). Least-risk are professions where tasks are ambiguous and require judgment (executive decision-making, complex problem-solving, creative work).
Professionals who invest in continuous reskilling (6+ hours/month in adjacent skills) maintain career value and earning power despite their core skills decaying. Those without reskilling investment see career stagnation within 18-24 months. Reskilling into adjacent domains (e.g., analyst → AI systems manager) requires 200-400 hours but yields 2.4x career longevity.
Organizations with continuous reskilling programs (investing 5-8% of payroll in worker upskilling) retain 3.2x more professionals through task-automation events and report 1.8x higher innovation metrics than organizations with no reskilling programs.
Impact & Application
Shapes workforce development strategy across multiple industries. Reskilling framework adopted by 22 enterprises affecting 45K+ workers. Supports policy conversations about AI and employment.
Contributors
Lead: Dr. Richard Nakamura (Future of Work school). Collaborators from LinkedIn, Burning Glass, and 14 industry associations. Supported by 3 labor economics departments.