Every distinct HR job ad on LinkedIn across the United States and 20 European countries, posted in the last seven days, split into five role families. Duplicate postings are stripped out. Every match was read in full and classified by how the ad uses AI.
One in twenty HR job ads makes AI a hard requirement, 182 of 3,371. Nearly one in five mentions AI somewhere in the text, and one in ten asks the candidate for any AI skill at all. Most of the talk carries no expectation.
European ads mention AI more often than American ones, 21% against 17%. The gap is narrower for asking the candidate to have the skill, 10% against 9%. Learning and development asks for AI skill more than any other HR role, and European L&D ads mention AI at twice the American rate.
Where ads name the skill, they mean the everyday assistants, Claude, ChatGPT, Copilot and Gemini, applied to the profession's own workflow. Machine-learning depth appears in under 2% of the skill-demanding ads.
Whether ads mention AI at all, and how much of that talk places any expectation on the candidate.
The green bar is the share of each family's ads that mention AI anywhere. The orange bar is the share that asks the candidate for AI skill. Recruiters talk the most, learning and development asks the most, and the distance between the bars is employer talk that expects nothing of you. Hover a family for the counts.
Each family's AI mentions broken down by type. The recruiter bar is the most diluted, 96 of its 246 mentions are about the employer rather than the candidate. The L&D bar is the densest in real expectations.
A further 73 ads are excluded from every figure on this page because AI appears only in legal boilerplate about the application process, typically a line stating whether AI screens applications. That says something about legal departments and nothing about the job. The split was 23 recruiter ads, 17 generalist, 16 L&D, 15 business partner and 2 leadership.
The 426 ads that ask for AI skill, or where the job itself uses AI, coded by capability, phrasing and named tools.
Coded against a fixed capability taxonomy. An ad can ask for more than one. Read down a column to see a profession's profile, across a row to see who wants that capability. Hover any cell for the counts.
The most common way each profession phrases the ability it wants, summarised from the ads. Recruiters want AI pointed at the funnel, L&D wants it pointed at content and at teaching others, generalists want their manual work to disappear, and leadership wants someone who can run the change.
AI products named in the 426 ads that ask for AI capability. HR systems like Workday and Greenhouse are excluded, since ads name them as the platform the work sits in rather than as AI skill. The four general assistants dominate, and the HR-specific AI tools barely register. Employers expect fluency with the same assistants everyone else uses, not with a niche HR stack.
One from each of the four families that ask for AI skill most often, all of them hard requirements. The full set of 630 AI-mentioning ads, filterable by role and by how AI is mentioned, sits on its own page.
The right candidate uses AI and data to accelerate design, improve access, and drive better outcomes, while applying strong judgment about accuracy, applicability, and where human expertise matters most.Open the ad →
Previous experience moving products beyond incremental improvement to reimagine end-to-end journeys, using technology, automation and AI where appropriate to improve speed, quality and user experience.Open the ad →
Ability to use HR data, workforce analytics, HRIS platforms, AI-enabled insights, and business metrics to influence decisions while protecting employee privacy and ensuring appropriate governance.Open the ad →
Experience leading in a data-rich, AI-enabled environment, using analytics and digital tools to support sound decisions, strengthen workforce outcomes, and promote responsible innovation.Open the ad →
The same measures cut by market and by country.
European ads mention AI more often overall, 20.9% against 17.1%. Business partners are the only family where the American rate is higher. European learning and development ads mention AI at more than twice the American rate, and the generalist gap runs the same way.
All five families pooled. Countries with at least 20 ads in the window. France, the United Kingdom, Poland and Spain form a top cluster at around a quarter to a third of ads, well above the United States. Italy, Belgium and Austria sit below one in ten. The orange inner bar is the share that asks for AI skill, which is where France and Spain separate from Poland and Portugal. Faded rows have fewer than forty ads and are indicative only.
| Country | Ads | Mention AI | Share | Ask for the skill |
|---|---|---|---|---|
| France | 81 | 27 | 33.3% | 12 |
| United Kingdom | 419 | 117 | 27.9% | 60 |
| Poland | 52 | 14 | 26.9% | 4 |
| Spain | 88 | 23 | 26.1% | 13 |
| Germany | 220 | 44 | 20.0% | 27 |
| Portugal | 48 | 9 | 18.8% | 2 |
| United States | 1,977 | 339 | 17.1% | 178 |
| Ireland | 35 | 5 | 14.3% | 3 |
| Sweden | 22 | 3 | 13.6% | 1 |
| Switzerland | 43 | 5 | 11.6% | 2 |
| Netherlands | 104 | 12 | 11.5% | 6 |
| Greece | 28 | 3 | 10.7% | 1 |
| Romania | 29 | 3 | 10.3% | 2 |
| Italy | 79 | 7 | 8.9% | 1 |
| Belgium | 59 | 4 | 6.8% | 2 |
| Austria | 31 | 2 | 6.5% | 2 |
Source. LinkedIn job listings pulled on 2 and 3 August 2026, covering ads posted in the preceding seven days. See the note on the two sources below.
Role families. HR leadership covers CHRO, Chief People Officer, VP HR, VP People, Head of HR, Head of People and HR Director. Business partners cover HR Business Partner, HRBP, People Partner and People Business Partner. Generalists cover HR Generalist, HR Specialist, People Operations, HR Advisor and HR Officer. Recruiters cover Recruiter, Talent Acquisition in all its variants and Sourcer. Learning and development covers Learning & Development, Instructional Designer, Training Manager, Training Specialist and Talent Development. Where a title matched two families, for instance Head of People Operations, it was assigned to the more senior one.
Two sources. The three HR families were pulled from LinkedIn's job index through the Apify actor fantastic-jobs/advanced-linkedin-job-search-api, which offers a server-side full-text filter and a built-in staffing-agency exclusion. Recruiters and learning and development were pulled later, through Monid's LinkedIn job search, because the first source had hit its monthly quota. That second source has no agency filter, so agencies were removed here by employer name and industry, which is a blunter instrument. It also has no server-side text filter, so the full ad body was pulled and matched locally against the AI term list before classification. Treat comparisons between the two groups as directional rather than exact, particularly for recruiters, where residual agency listings are the likeliest source of error.
Geography. The United States plus the United Kingdom, Germany, France, the Netherlands, Sweden, Denmark, Norway, Finland, Spain, Italy, Ireland, Belgium, Poland, Switzerland, Austria, Portugal, Czechia, Romania, Hungary and Greece.
Duplicates are gone. Employers routinely publish the same opening across many locations. Repeat postings were dropped from the dataset before any analysis, leaving 3,371 distinct combinations of employer and job title. Every number on this page counts an opening once, no matter how many times LinkedIn carried it.
Why seven days. Seven days answers the question of what is live right now, in one clean snapshot. A thirty-day pull would cost roughly four times as much, mix expired ads with live ones, and change the shares very little.
Detection runs in sixteen languages. Every one of the 3,371 ad bodies was matched locally against an AI vocabulary covering English, German, French, Spanish, Portuguese, Italian, Dutch, Swedish, Danish, Norwegian, Finnish, Polish, Czech, Romanian, Hungarian and Greek, including the acronyms KI, IA and AI, and the named assistants. Acronyms are matched case-sensitively and only in the language regions where they mean artificial intelligence, which keeps the Italian preposition "ai" and the French "j'ai" out. That pass recovered 44 ads the original English-only search had missed, concentrated in Germany, France and Spain.
Classification. Every match was read in its entirety and classified against a fixed category set, then quality-checked in an independent pass that verified each quote appears verbatim in the original and that the category holds. A further 73 ads were set aside as application-process boilerplate, and thirty-five matches were discarded as false positives, among them the Italian preposition "ai", the French "j'ai", company codes containing IA, scraped page furniture, and the string "anthropic" hiding inside "philanthropic".
Where the numbers still strain. The language gap is closed but two limits remain. Countries below roughly forty ads, Ireland, Austria, Romania, Greece and Sweden, carry wide error bars and should be read as directional rather than precise. And an ad that gestures at AI without naming it in any of the sixteen languages, for instance by describing automation or intelligent tooling in the abstract, is still invisible to this method. The United States, the United Kingdom and Germany are the three steadiest series.
Every number on this page, plus the underlying ads, as one markdown file. Open it in any editor, drop it into a spreadsheet, or hand it to a model and ask your own questions of it.
Markdown, 619 KB. Free to reuse with a link back to this page.