The half-life of a professional skill has fallen from over a decade in the 1980s to four or five years today, and to roughly two to two and a half years for technical and AI skills.1 The World Economic Forum estimates that the larger share of AI's projected $15 trillion economic prize will be captured through workforce learning rather than the technology itself.1 By 2025, close to half the global workforce required significant reskilling to stay current. Learning has moved from a support function to a performance constraint. Skill half-life has collapsed to two years. Output is now gated by how fast your team absorbs new tools and ways of working — not by how much training it logs.
Why It Matters
For a functional leader, this changes what you are actually managing. Output in an AI-augmented team is now gated by how quickly people can absorb a new tool, a new model, or a new way of working and apply it to real work. In 2026, leading organisations have started measuring this directly, through metrics such as "capability velocity" — the speed at which a team builds the skills a new strategy needs. Training hours logged is the wrong number to optimise.
There is a cost to ignoring this. The fastest-growing source of workforce anxiety in 2026 is the fear that skills are decaying faster than they can be rebuilt. That anxiety shows up as slower adoption, quiet resistance, and the loss of exactly the people you most need to keep. Workers with current AI skills already command around a 56% pay premium in the same roles,2 so the market is repricing learning faster than most internal pay and development systems are.
The deeper shift is who owns learning. When skills decayed over a decade, training could sit with a central function and run on an annual cycle. At a two-year half-life, learning has to live inside the work itself, owned by the team lead and measured like any other output. The functional leaders pulling ahead have stopped outsourcing capability to a training calendar and started treating it as part of how the team operates week to week.
6xD Interpretation
- Primary lens — D5: Digital Worker & Workspace. A collapsed skill half-life redefines what the digital worker's job actually is: continuous absorption of new tools and methods, not a fixed competency set.
- Supporting lens — D2: Digital Cognitive Organization. Learning owned by the team and measured week to week is an organisational sensing-and-adapting capability, not a training-calendar event.
- Supporting lens — D1: Economy 4.0. The wage premium for current AI skills is the labour market repricing capability in real time — a structural signal, not a compensation footnote.
6xD Insights interpretation: At a two-year skill half-life, learning velocity is not a support metric. It is the constraint that decides whether the rest of the operating model can keep pace with the work.
Executive Implications
| Decision area | Executive question | Required output |
|---|---|---|
| Learning ownership | Does learning sit with a central training function, or with the team lead? | Learning owned and measured at the team level |
| Capability velocity | Is the organisation tracking how fast teams absorb new capability? | A tracked "capability velocity" or equivalent metric |
| Compensation alignment | Does pay reflect current, not historical, skill value? | Pay practices reviewed against the AI-skills wage premium |
| Cadence | Is training continuous or an annual event? | Short, continuous learning cycles built into the work |
Recommended Actions
- Put a number on learning speed. Pick one capability your team will need within the year, and measure the time from "we should learn this" to "we are using it in the work."
- Move ownership to the team. Stop treating capability as a central training-calendar responsibility; make the team lead accountable for it.
- Set a target this planning cycle. Hold the function to a defined learning-velocity target rather than a training-hours quota.
- Check pay against the market. Review whether current compensation reflects the wage premium the market already places on AI-relevant skills.
Sources
- 01World Economic Forum workforce-reskilling and AI economic-value estimates; industry skill-half-life analyses (2026). ↩
- 02[PwC, "5 takeaways from the 2025 AI Jobs Barometer" (3 June 2025)](https://www.pwc.com/us/en/tech-effect/ai-analytics/ai-jobs-barometer.html). ↩
