Explore intelligence
Selected Hong Kong SAR, China. Map lens AI conditions, value 93, Stronger conditions.

humAIne intelligence · Global AI Atlas

What should this country ask
before it trusts AI with real lives?

Reading the atlas: the context map is a directional model. Adoption estimates and surveys are shown separately in each country profile, with their actual dates. Judgement stays human.

Country context, AI adoption and social impact in one atlas. Explore the conditions for useful AI alongside evidence about how people and organisations use it.

Search a country for its opportunities, adoption estimates and social-use surveys. Compare like-for-like measures, inspect the original sources and explore economic studies. Each measure keeps its own population and observation period.

Quarterly from 1 October 2026: 1 January, 1 April, 1 July and 1 October at 08:00 Europe/Zurich.

Read the method

Selected

Hong Kong SAR, China

AI conditions: 93 · Stronger conditions

Scale with safeguards

Vulnerable work

5.8%

Unemployment

2.8%

Inflation

1.4%

Countries first

Search and select below. The map is secondary on small screens.

AI conditions

Access, skills, public capacity and resource room

LowerHigher

Directional model. Country comparison guides questions; it does not replace local diagnosis.

Selected country indicators and lens values
IndicatorValueYear
AI conditions93Stronger conditions
Vulnerable work5.8%2025
Unemployment2.8%2025
Inflation1.4%2025
Internet use96%2024
Electricity access100%2024
Secondary completion129%2024
Life expectancy85.4 years2024
Population 65+24%2025
Water stressNo recent value—
Public capacity83 / 1002024

Vulnerable work

5.8%

Observation 2025

Unemployment

2.8%

Observation 2025

Inflation

1.4%

Observation 2025

Internet use

96%

Observation 2024

Electricity access

100%

Observation 2024

The FII survey covers 60,000 people in 32 countries, not every country on this map. Regional worry patterns therefore act as a starting lens. Country indicators adjust the analysis and every panel shows where evidence is thin. One human remains responsible for judgement; this is a research tool under Intelligence, not a firm mandate.

217Country context profiles
14Comparable indicators
219Adoption & social profiles
4×Reviews per year

Global AI Atlas / Evidence edition 2026-08.r2

Country context checked 2026-09-01. Adoption and social evidence checked 2026-09-07. Source observation dates remain attached to every value. The source registries have different coverage; evidence-only economies are available in Adoption & society.

Economic evidence / native study periods

What happens when people use AI?

Different tasks, different results. These studies are outside the country ranking.

Randomized field experiment

Consulting tasks and the limits of assistance

Over 25% faster on tasks within GPT-4’s capability

758 BCG consultants completing realistic knowledge-work tasks. The reported speed gain applies to the tasks within the tested model’s capabilities.

Working paper reported 21 September 2023 · journal publication March 2026

Performance depended on the task; assistance could worsen results beyond the model’s capabilities. This is historical, task-specific evidence, not a national productivity estimate or a claim about today’s models.

Read the study ↗ · 2026 publication ↗
Randomized controlled study

Experienced developers, familiar repositories

AI-allowed tasks took 19% longer

16 experienced open-source developers; 246 issues in repositories they knew.

Early 2025 experiment · published July 2025

This setting does not represent all developers or current tools. METR’s February 2026 follow-up identifies selection effects and redesigns the study; do not extrapolate this effect to countries.

Read the study ↗ · 2026 publication ↗
Observational study of staggered adoption

Customer support in one organization

15% more issues resolved per hour

5,172 customer-support agents in the authors’ updated paper.

Study published 2025 · earlier company rollout

Effects vary by experience and skill. A company-specific productivity result is not measured national output or realized net savings.

Read the study ↗
Explore adoption & society →Sources & coverage →Evidence editions →

Countries and economies follow source reporting conventions; boundaries and listings imply no position on sovereignty. Missing polygons remain accessible through country search and the table. AI estimates, surveys and studies are separate evidence. No universal impact score is calculated.
Eurostat data selected and reformatted by humAIne with attribution. Microsoft data © 2026 Microsoft, reused under the MIT License. Interpretation is humAIne’s responsibility.

Seven regional questions

A shared technology.
Different questions.

The regional layer comes from the worry patterns in the FII Priority Compass and is refined with country indicators. It frames inquiry, not a playbook. Every country still needs local evidence and public participation — and judgement stays with a named human.

Method and limits

Evidence first.
Judgement stays human.

The atlas is a decision aid for asking better questions. It does not forecast national outcomes, replace local diagnosis, or recommend automated decisions about people.

Regional worry lens

FII Priority Compass 2025 establishes the public concern pattern. Its 32-country sample supports regional context, not direct claims about every country.

Country evidence

Fourteen World Bank series add labour, prices, access, health, education, demographics, water and public-capacity signals. Each panel keeps the observation year.

Directional model

Human pressure combines measured constraints. Deployment conditions combine access, skills, public capacity and resource room. The bridge lens joins the two.

Rights boundary

UNESCO principles guide the guardrails: proportionality, privacy, auditability, transparency, human responsibility and routes to challenge decisions.

See the model formula and coverage limits

Human pressure

Available signals include vulnerable work, unemployment, inflation, internet and electricity gaps, life expectancy, secondary completion, water stress and government effectiveness. Current FCV classifications add a conflict and fragility signal.

Deployment conditions

Access averages internet and electricity. Skills use lower-secondary completion. Public capacity uses the revised World Bank score. Resource room uses GDP per person at purchasing-power parity on a log scale.

Bridge gap

The bridge lens weights human pressure at 62% and the inverse of deployment conditions at 38%. Missing observations are omitted or replaced by a stated fallback. Bands reduce false precision.

Evidence base · Checked 2026-09-01

Open sources.
Visible limits.

Indicator years vary because international series update on different schedules. Use the country panel year before drawing a conclusion.

01FII Priority Compass 2025Global and regional public worries, 60,000 respondents across 32 countries02World Bank Data APICountry metadata and the 14 indicator series used in the model03World Bank AI Foundations 2025Connectivity, compute, context and competency as foundations for AI participation04ITU Facts and Figures 2025Connectivity, affordability, quality and digital-skills divides05ILO World Employment and Social OutlookLabour-market, youth-employment and job-quality context06UNDP Human Development Report 2025People, agency and choices in the age of AI07UNESCO Recommendation on AI EthicsHuman rights, oversight, transparency, privacy and accountability08World Bank FY2027 FCV classificationsCurrent conflict exposure and institutional fragility lists09SVG World MapCountry geometry, CC BY 4.0, based on MapSVG

Responsible AI starts with the public outcome, local evidence and a named human who remains accountable.

Return to the map
Global AI Atlas | humAIne