Methodology

The outcome that matters

The aim is to assess vulnerability to losing a job because of AI or automation, or to an involuntary reduction in paid hours that leaves earnings insufficient to live on. Faster tasks and hours saved are intermediate evidence. They do not directly measure this outcome.

A supported scenario must account for local pay and living costs, involuntary job or hours losses, and opportunities to find other paid work. No livelihood-risk model or numerical category thresholds have yet been approved. Missing assessments remain explicit.

Indicative scenarios for work and learning

Indicative GenAI scenario with site assumptions. It does not predict personal dismissal; robotics and other automation are not included.

The 147-economy release uses the same international Microsoft GenAI diffusion measure everywhere: estimated use for any purpose among ages 15–64 in June 2026. It does not mix this population estimate with workplace-only Eurostat survey rates.

ILO–NASK WP140 (2025) provides technical task exposure of the broader ISCO group. Both the group and the population adoption measure are proxies for a detailed occupation. Microsoft telemetry is adjusted for device share, internet access and population; coverage of local AI products can be incomplete, notably in China.

From the declared June 30, 2026 baseline anchor to the end of 2030 or 2035: f=2^((t−2026+0.5)/d), A=a·f/(1−a+a·f), index=100·E·A. E is the ILO index, a is the population-use estimate, and d is 3 years with 5/2-year alternatives. Anchors, growth rates and thresholds are site assumptions, not source forecasts.

Five fixed editorial bands: below 5, [5,15), [15,30), [30,50), and 50 or above. We check growth alternatives and all thresholds shifted by ±20%. This is an internal comparison index, not percentages of layoffs, displaced hours or a scientific confidence interval.

Full catalogs and source-specific gaps remain. This index covers GenAI only; other channels are shown separately when available. Actual paid-hour losses, future pay and competition are unknown. Career suggestions compare skills and the target’s own same-country/horizon scenario. The editorial model is self-reviewed; no independent scientific validation is claimed.

Career exploration by skills

The full ESCO graph keeps essential and optional skill relationships. Options are ordered by the share of target essential skills covered, then by fewer gaps and more shared skills. This is classification overlap, not a chance of employment. The default source skills are an assumption you can edit. Personal filters are not included in shared links.

Historical EURES shortage/surplus lists describe wider ISCO groups and an incomplete source table. Unlisted statuses are unknown; opposing regional reports are retained. Current regulatory notes are scoped to the named profession and territory. They do not establish individual qualification recognition, language proficiency, a training route or the right to work. Future pressure and competition are assessed separately; unknown is not lower.

Entry-pathway coverage

Official National Careers Service source for England. The complete index of737 source titles was matched against all3039 catalog occupations.

Entry-pathway lists: 324; unambiguously linked ESCO occupations: 306. 2733 occupations lack a usable unambiguous link; this source is not assigned to262 other geographies. The full catalog is retained.

These are current entry pathways and study-subject examples. They do not establish personal eligibility, duration, cost or program availability in2030/2035.

Map view coverage

Each view retains the full catalog. Combined estimates require both components; absence of one is not low risk.

The combined index is the larger prepared AI or robotics index. Multiple occupation mappings use the upper alternative for colour, with the range retained. Fixed editorial cutpoints: 5, 15, 30, 50; adoption alternatives and ±20% band shifts are checked. These guides support comparison and learning exploration, not dismissal probabilities.

Physical-task context

Published robot capability scores, US 2026. Official ESCO–O*NET URI mapping was checked for every catalog occupation.

With source context: 2471 / 3039. Explicit mapping/source gaps: 568. Native O*NET scores: 923.

This historical layer does not add country-year forecasts. The map shows the separate GenAI outlook. Where available, robotics scenarios are counted separately below; total automation risk and actual hour losses are not established.