PwC Analysed One Billion Job Ads — and the Answer Is Not What Most Boards Expected

AI is not primarily a job destroyer or a job creator. It is a divider — splitting the global labour market into two tracks with fundamentally different growth rates, wage trajectories, and skill requirements. Most workforce strategies were not designed for that finding.

Sign up for my Substack daily AI newsletter here.

See my AI Training course portfolio for corporate Business Leaders here.

Article link: https://open.substack.com/pub/johanosteyn/p/pwc-analysed-one-billion-job-ads?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

On 15 June 2026, PwC released its 2026 Global AI Jobs Barometer — the most comprehensive empirical study of AI’s impact on the labour market published to date. The methodology is significant: more than one billion job advertisements analysed across 27 countries and six continents, combined with company financial data and occupational task data. This is not a forecast. It is not a survey of executive intentions or expert predictions. It is an analysis of what employers have actually been asking for, paying for, and hiring for — at a scale that eliminates most of the noise that distorts smaller studies.

The headline finding is both precise and counterintuitive. AI is not primarily eliminating jobs or creating them. It is dividing the labour market into two fundamentally different tracks — and the distance between those tracks is growing faster than most workforce strategies have accounted for.

CONTEXT AND BACKGROUND

PwC’s framework divides AI’s labour market impact into two categories that deserve careful attention. The first is professionalised roles — occupations in which AI automates routine tasks, amplifying human expertise and making the remaining human judgment more valuable. Radiologists whose AI handles image screening so they can focus on complex diagnoses. Recruiters whose AI processes initial screening so they can conduct more meaningful conversations with candidates. Analysts whose AI handles data aggregation so they can spend more time on interpretation and strategy. The second is democratised roles — occupations in which AI makes the role itself easier to perform, reducing the expertise required to do the work and therefore reducing the premium the labour market places on it. Medical secretaries whose documentation work AI now handles. IT service managers whose routine troubleshooting AI now resolves. Administrative roles whose value was largely in their volume and their knowledge of specific systems, both of which AI can now replicate.

The labour market consequences of that distinction are already measurable and significant. Professionalised roles are seeing 39 per cent growth in job availability since the 2018 baseline, compared to 17 per cent growth for democratised roles — roughly twice the rate — and are recording 42 per cent faster salary growth. The companies operating in the most AI-exposed sectors recorded 34 per cent productivity growth in 2025 relative to 2018, compared to 24 per cent for the least AI-exposed companies. Within this group, a super-star effect is emerging — the top 20 per cent of most AI-exposed companies achieved average labour productivity growth of 163 per cent relative to 2018, nearly five times higher than the broader most AI-exposed group.

Perhaps most surprisingly, headcount growth at the most AI-exposed companies is outpacing growth at the least AI-exposed companies — 52 per cent relative to 36 per cent since the 2018 baseline. This is the finding that most directly contradicts the dominant narrative of AI as a workforce reduction technology. The organisations using AI most effectively are not shrinking their workforces. They are growing them faster than their less AI-capable peers — and growing them toward roles that require more human expertise, higher human judgment, and more specifically human skills.

INSIGHT AND ANALYSIS

The data point that carries the most strategic weight for boards and executives is the entry-level finding. Based on 2.4 million entry-level jobs analysed in the US, entry-level roles most exposed to AI are now seven times more likely to require traditionally senior-level human-intensive skills like leadership, creativity, and face-to-face interaction. Job openings for these seniorised entry-level roles have grown 35 per cent since 2019. Other entry-level roles shrank by 10 per cent over the same period.

PwC Global Workforce Leader Pete Brown names the structural consequence of that data precisely: “AI is removing some of the routine work that once acted as an apprenticeship, while increasing demand for judgement, leadership and adaptability much earlier in careers. Organisations need to rethink how they develop talent if they want people to thrive in this new environment.”

The apprenticeship observation is the one that most current workforce strategies have not yet absorbed. The professional development pipeline that produced senior talent across accounting, law, consulting, financial services, and technology was built on junior people doing the volume cognitive work that AI is now automating. Audit associates reviewed documents. Graduate analysts built financial models. Junior developers wrote boilerplate code. Junior lawyers conducted initial research. Those roles were not only sources of billing margin or production output. They were the mechanism through which organisations transferred professional judgment, domain knowledge, and institutional context to the next generation of senior practitioners. When AI automates the work, the billing margin disappears first. The judgment transfer pipeline disappears second — and it disappears more quietly, accumulating as a capability deficit that will become visible in the senior ranks years later.

I have previously written about the generation that chose their degrees for a labour market that no longer exists — the specific cohort of students who entered university programmes in fields now most exposed to AI automation, and who are discovering that the entry-level roles those degrees were supposed to unlock are being eliminated faster than the market can create alternatives.

The PwC data contextualises that individual-level observation within a structural labour market finding: the entry-level roles that are disappearing are not the ones that were going to produce AI-capable senior professionals. The ones that are growing — the seniorised entry-level roles that require leadership and judgment from day one — are the ones that require a fundamentally different preparation than the one most university programmes currently provide.

The wage premium data requires careful reading to extract its full strategic implications. Jobs requiring specific AI skills — prompt engineering, machine learning, AI system design and deployment — are growing roughly eight times faster than the total jobs market. The global average wage premium for workers with those skills has risen to 62 per cent, up from 57 per cent the previous year. That premium, however, diverges sharply by sector: employers in consumer markets are paying a 118 per cent premium for AI talent, while the premium in government and public sector work sits at just 16 per cent. These are wage premium figures, not job growth rates — and the distance between them describes a structural talent allocation problem that has consequences well beyond compensation benchmarking.

IMPLICATIONS

The two-track labour market has implications for South African boards and executives that extend and sharpen the global findings of the PwC report.

The first is about AI strategy design. PwC Global Chief AI Officer Joe Atkinson’s framing is precise: “The companies seeing the greatest returns on AI are using it to amplify human expertise, accelerate innovation and create entirely new sources of value.” The companies on the lower track — using AI primarily to democratise roles, reduce the skill required to perform them, and extract cost savings from workforce reduction — are growing more slowly, paying less, and falling further behind on productivity. For boards approving AI strategies, the question of which track the organisation’s AI investment is placing it on is the most important strategic question the PwC data raises. The answer is not always in the strategy document. It is in the specific deployment choices being made at the operational level.

The second is about talent development. The organisations that will be most capable in five years are the ones that are redesigning their early-career development models now for a world where the apprenticeship has been partially automated. That redesign requires a clear answer to the question of what junior roles should look like when AI handles the volume cognitive work — what the judgment-building, knowledge-transfer, and professional socialisation functions of those roles should be in a world where the task content that used to carry them has changed. Most organisations have not yet had that conversation.

The third implication is the one most specific to South Africa — and the most urgent. The 16 per cent public sector wage premium for AI skills against 118 per cent in the private sector is not simply a talent migration pressure. It is an existential threat to South African state capacity. South Africa already confronts well-documented governance challenges in municipal delivery, state-owned enterprise management, regulatory oversight, and the administration of public institutions whose effectiveness determines whether the country’s legal, financial, and social infrastructure functions adequately. The AI-capable data scientists, legal technology experts, regulatory analysts, and systems designers that public institutions need to govern AI-driven commercial environments — to enforce POPIA, to regulate agentic financial systems, to audit AI-assisted public procurement — will not remain in public service at a 16 per cent premium when private sector employers are offering 118 per cent. The result is not only a talent gap inside government. It is a regulatory capacity gap that compounds every year it is not addressed — a state that becomes progressively less able to govern the technologies its private sector is deploying, at precisely the moment that the PwC data shows those technologies are reshaping the economy at the fastest pace in recorded labour market history.

CLOSING TAKEAWAY

The debate about AI and jobs has been dominated by competing forecasts — millions of jobs created, millions destroyed, net positive, net negative — presented with confidence that the data has never justified. PwC’s one-billion-job-ad study does not resolve that debate. What it does is replace speculation with evidence about what is actually happening in labour markets right now. And what is happening is not primarily creation or destruction. It is divergence.

The organisations on the professionalised track — using AI to amplify human expertise, growing headcount faster, paying more, producing dramatically higher productivity — are pulling ahead. The organisations on the democratised track are moving more slowly. The distance between them is measurable, compounding, and already significant. For South Africa, the divergence carries an additional dimension that the global report does not fully name: a public sector progressively hollowed of AI capability, unable to govern the technologies reshaping the private economy, in a country where the quality of governance determines whether any of the productivity gains the PwC data describes reach the people who need them most.

Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net