Will AI Take Your Job? The Two-Track Job Market

The short answer

Not in bulk, and not in the way the headlines describe it. According to the widest study of job advertisements published so far, AI is not deleting the job market. It is splitting it in two, and the two halves are moving in opposite directions.

For its 2026 Global AI Jobs Barometer, PwC analysed more than one billion job advertisements across six continents and 27 countries and territories. The result is three groups. 22% of jobs are being professionalised: AI absorbs the routine work and leaves the harder judgement, so the role demands more expertise than before. 52% are being democratised: AI absorbs the expert tasks and leaves the simpler ones, so the role demands less. The remaining 26% have low exposure, where AI touches few of the tasks.

The professionalised track is pulling ahead in numbers, in complexity and in pay. That is the finding worth acting on, and it is rarely the one quoted.

Read it with two things in mind. The report measures what employers ask for in job ads, not what happens to the people who fill them. And its author is a consultancy that sells AI transformation, which is not a reason to dismiss it, but is a reason to say it out loud.

The two tracks, side by side

The posting and salary figures are the global series, from the six countries where comparable data runs back to 2012. The difference in salary growth between the two tracks is 42% in relative terms.

Why the professionalised track is pulling ahead

Three numbers carry the divergence.

Job postings for professionalised roles grew 39% between 2018 and 2025; for democratised roles, 17%. Both tracks are still growing, but one is growing twice as fast.

Salaries moved the same way. Professionalised advertised salaries grew 37% since 2021, democratised salaries 26%, and the gap has widened since 2022, the year AI use accelerated.

And the roles became harder to fake. The number of skills employers demand in professionalised jobs grew 68%; in democratised jobs, 33%. Complexity is being added on one track and removed from the other.

Your job title does not tell you which track you are on

The most common misreading is taking the split as a list of safe jobs and doomed jobs. It is not that, and the report's own examples show why.

An employment recruiter is on the professionalised track. Screening CVs used to be where a junior learnt the trade; AI does that now. What is left for the person is contract negotiation and the judgement calls around it. Same title, more expertise required.

A software developer sits in the democratised group. AI now performs part of the expert work of writing code, so the role is pushed toward the simpler tasks around it. Same title, less expertise required.

So the useful question is not whether your profession appears on a list. It is this: once AI takes its share of my tasks, is what remains harder or easier than what I do today? Two people with the same title, in two companies that designed the role differently, can be on opposite tracks.

The same split appears at the entry level

If you are early in your career, this is the part that concerns you most, and it is the part with the clearest evidence.

AI-exposed junior roles are not all shrinking. They are being seniorised. In the US data, entry-level roles most exposed to AI are seven times more likely to require traditionally senior skills than the least exposed ones: motivational leadership, team building, stakeholder management, mentorship, data-driven decision making. For the most exposed roles, 52% of the newly required skills are of that kind. For the least exposed, 7%.

The entry-level roles that were seniorised in this way grew 35% between 2019 and 2025. The other entry-level roles declined 10%. In parallel, 49% of chief executives told PwC they expect AI adoption to reduce junior hiring over the next three years, against 12% who expect it to reduce senior hiring. A Stanford University analysis cited in the report found a 16% decline in entry-level jobs in highly AI-exposed fields.

None of that says AI is the cause. The report is explicit that other forces may be at work. It says something narrower and more useful: at the entry level, the roles that hold are the ones that now ask for more than they used to.

Four questions that place a role on a track

The report offers four questions for reading the future of one specific job. Together they are a checklist you can run before your next application.

1. Expertise. How is AI changing the level of expertise the role requires? If the routine goes and the judgement stays, you are on the professionalised side. If the judgement goes and the routine stays, you are not. 2. Supply and demand. How might demand for the role change, and how many people will be able to do it? A role can be democratised and still grow. Child care services managers, a democratised example, saw job listings more than double since 2019 while wages grew 8%. More people can do the job, the job grows, the pay does not. 3. The limits of AI. Where is a person still needed to check the output or handle the atypical case? That is where the remaining expertise sits, and where you want to be. 4. Outside forces. What constrains how much AI can actually be used here, from regulation to workflow? These are the roles where the model predicts one thing and the labour market does another.

What to do, track by track

Professionalised. The direction is in your favour, and the risk is complacency. Build the expertise the role now demands and be able to evidence it: the negotiation you led, the case you built, the decision nobody else was positioned to make.

Democratised. The title may stay common while the premium disappears. Move toward the judgement end of your own job, or toward an adjacent role where expertise is being added rather than removed.

Low exposure. AI is not the threat; the rebalancing around you is. The expertise you already have becomes more valuable if you can point it at work AI assists.

Early career. Do not wait for the years of drudgery to earn the senior skills. They are being asked for now. Leadership, judgement and stakeholder work are the entry ticket, not the promotion.

One more pointer from the same data, this time about employers. Headcount at the companies most exposed to AI grew 52% against 36% at the least exposed, with wages at 24% against 17%. The authors' reading is that the companies pulling ahead are using AI to redesign work, not only to trim it. When you are choosing between two offers, that difference is worth a question in the interview.

What a track does not decide

None of this makes the machine the decision-maker. The two-track finding is about the demand for expertise, not about automatic rejection, and the difference matters when you are writing your CV. If the system does not extract your dates, titles and skills, the person searching it finds nothing.

For the mechanics of that, and for the myth that most CVs are rejected by software before a human sees them, the guide below covers it.

How Lemu Work fits

Lemu Work does the repetitive work so your time goes to the part that now decides the outcome.

• Upload once: your CV as a PDF, or your LinkedIn profile, becomes a structured profile.
• Live listings: roles that fit that profile, verified before you invest time.
• A pack per role: tailored CV, cover letter and a score against your experience. You approve, you send.
• A growth plan: the training, certifications and adjacent roles that move you toward the expertise AI leaves in place.

It will not tell you which jobs will exist in five years, and no honest tool will. What it removes is the administration: reading listings, adapting documents, keeping the record of what you sent.

When you get hired, export your data and uninstall.

Try it

Open the Lemu Work app and run one role you are considering through the flow. Then read your own CV and ask the expertise question: after AI takes its share of these tasks, is what remains harder or easier? If most of what is left could be done by a machine, you have found both the roles to avoid and the training to start.

Sources

• PwC, 2026 Global AI Jobs Barometer, press release of 15 June 2026 (https://www.pwc.com/gx/en/news-room/press-releases/2026/pwc-2026-ai-jobs-barometer.html) and the full report (https://www.pwc.com/gx/en/issues/artificial-intelligence/job-barometer/2026/2026-global-ai-jobs-barometer-full-report.pdf): more than one billion job advertisements across six continents and 27 countries and territories
• Same report: the professionalised and democratised categories, the job posting, salary and skills series, the entry-level seniorisation data, and the Global CEO Survey figures on junior hiring
• Brynjolfsson et al., Stanford University, 2025, on the 16% decline in entry-level jobs in AI-exposed fields, cited in the report
• The four questions for reading a specific role, and the child care services managers example, as published in the report

Related Reading

• AI job search: a practical guide
• Does AI reject your CV before a human reads it?
• CV tailoring with AI
• Interview prep linked to the application you actually sent
• Ghost jobs and stale listings
• Europass or ATS format in the EU market?
• How to verify a job listing is still open
• AI job search in Portugal: the guide to the local market