SOURCES JobTech · Eurostat · AI Index · SCB · EU-LFS · Akavia8.1M DISTINCT SWEDISH ADS · 36 COUNTRIES● LIVE FEED 30 Sep 2026 · PUBLIC + PARTNER DATASOURCES CHECKED 28 Sep 2026 · SERIES LAST MOVED 30 Sep 2026MONITOR VERSION 1
A multi-country, multi-disciplinary research lab

We study how artificial intelligence is reshaping the world of work.

An economics-led, multi-disciplinary research lab at Örebro University and RATIO, founded in 2019. Örebro is a node of AISCAF, the WASP-HS research cluster. We combine administrative registers from several European countries with job advertisements, surveys and public cross-country data. The AIEL Monitor is where part of that work becomes public: open indicators on AI and work across countries, with Sweden in depth, updated as the data arrive.

Örebro University · RATIOWASP-HS · AISCAFSweden · Denmark · Portugal · Germany
AI exposure · share of jobs in the most exposed occupations 36 countries

39% of Swedish jobs sit in the most AI-exposed quarter of occupations, among the highest of 36 countries (mean 30%). Exposure marks where AI overlaps with the work, not what follows from it.

0%10%20%30%40%50%60%36-country 30.2Luxembourg56Sweden39Netherlands38Belgium38Switzerland36Germany35Cyprus35Denmark34United Kingdom ’1934France34Lithuania33Finland33
↓ Data (CSV)↓ SVGSource: DAIOE generative-AI v2023 × Eurostat EU-LFS employment 2023 · 36 countries, leading 12 shownMethod & sources →

All 36 countries →

What the lab is

A research lab first.

The monitor is our measurement infrastructure made public: the same data and measures our own research runs on. We work across countries on public data and international comparisons, and, rarely for any lab, on linked employer–employee register data in several of them: deepest in Sweden, reaching Denmark, Portugal and Germany, and expanding to more. Economists work alongside sociologists, business scholars and computer scientists, because a labour market changed by AI cannot be read from one discipline or one country alone.

01 · DATA

Register-grade evidence

Linked employer–employee register data at population scale in Sweden: annual, with 4-digit occupations (LISA), and monthly individual employment (AGI); plus comparable access in a handful of other countries. Rare reach, paired with 8.1M distinct public job ads (11.2M ad records).

02 · REACH

Multi-country

Register-level in Sweden, Denmark, Portugal and Germany, with more countries planned; 30-plus via EU-LFS and international job-ad data for external validity.

03 · LENS

Multi-disciplinary

Economics, sociology, business administration and computer science, partly through the WASP-HS cluster AISCAF, co-led with Uppsala and Stockholm.

04 · OUTPUT

Open public goods

Peer-reviewed research, plus citable, versioned public tools: the AIEL Monitor and the DAIOE exposure measure and Explorer.

Flagship · The AIEL Monitor

Sweden, in depth: four things the live job-ad data show.

One of the monitor's country cuts, read from Sweden's public job-ad stream (JobTech / Platsbanken). Every figure is measured from the ad text with a versioned, citable term list. Where something is not yet measured, we say so.

AI in Demand · share of Swedish job ads live

We measure AI demand in job ads as a range. The floor counts ads that ask for AI in the job's own requirements; the ceiling, our broadest measure of AI demand written in words, adds AI terms anywhere in the ad, and the one in eight ads that only say “AI” which turn out, read by hand, to be genuine AI roles. In 2025 AI demand was between 0.54% and 1.23% of ads; 2026 so far (January–June) runs higher, between 0.59% and 1.47% (provisional). Counting naively every ad that mentions AI in any form, a bare “AI” included, would bring the share to 2.42% in 2025 and 3.34% in 2026 so far; most of that difference is AI talk rather than demand. Ads naming an AI term anywhere, the thin middle line, reached 1.06% in 2025, 38 times the pooled 2006–2008 level, with generative-AI skills now 27% of them.

Ceiling (broadest measure) AI term named anywhere Floor: asks for AI in the role ╌ newest point provisional
↓ Data (CSV)↓ SVGSource: JobTech / Platsbanken job ads (CC0), 2006 onwards · frozen v1.6 term list · distinct advertisementsNext: tier split: built, integrated or simply usedMethod & sources →
0.54–1.23%
of 2025 job advertisements ask for AI skills, from the floor (asked of the person hired) to the ceiling (AI terms anywhere, plus the hand-read share of ads that only say “AI”).
Platsbanken · frozen v1.6 · distinct ads · 2025
38×
rise in ads naming an AI skill, from a pooled 2006–08 base to 2025 (0.028% → 1.06%, 95% interval 32–43×); the strict floor rose 41×. Measured instead from a 2015–17 base, where the term list has been stable across freezes, the rise is 6.3× (6.0–6.6).
Platsbanken · base 2006–08 (675,259 ads, 190 flagged) · 2025
28.8
ads per 10,000 named a generative-AI skill in 2025, from near zero before 2023; 27% of AI demand.
Platsbanken · 2025
−5.3pp
fewer entry-level openings in the most AI-exposed occupations, a gap widening since 2020; descriptive, and not separable from the rate cycle.
JobTech · 2020–2025
The Monitor · public data

The AI-Econ Lab Monitor

How AI shows up in the labour market: AI in Demand (live), the Occupations Explorer (live), AI exposure across countries (live), and modules on adoption, augmentation and barriers in development.

Open the monitor →
The measure · open & versioned

DAIOE

Our data-driven AI Occupational Exposure measure, published openly and mapped across SOC / ISCO / SSYK so others can join it onto their own data.

Explore DAIOE →

Research

Selected recent work.

2026Artificial Intelligence for Public UseLodefalk, M., Engberg, E., Lidskog, R. & A. TangDigital Society (accepted)
2025Artificial Intelligence, Tasks, Skills and Wages: Worker-Level Evidence from GermanyEngberg, E., Koch, M., Lodefalk, M. & S. SchroederResearch Policy (early online)
2025Artificial Intelligence, Hiring and Employment: Job Postings Evidence from SwedenEngberg, E., Hellsten, M., Javed, F., Lodefalk, M., Sabolova, R., Schroeder, S. & A. TangApplied Economics Letters (early online)
2025Artificial Intelligence and Worker Stress: Evidence from GermanyKoch, M. & M. LodefalkDigital Society, 4(5)

All 22 papers →