Predictions about the future of work tend to swing between extremes. Either AI will replace large portions of the workforce, or demographic decline will stall growth for decades. Both forces are real. Neither tells the full story on its own.

At Recruit and Indeed, we see millions of workers and employers interacting every day across 60 countries. Our Hiring Lab economists analyze job postings, searches and hiring patterns in real time. That vantage point provides a clear signal through the noise. The labour market is cooling, but it is not collapsing. And AI is transforming work, not eliminating it.

Demographics are the defining constraint

Across advanced economies, labour supply growth is slowing due to ageing populations and stricter immigration policies. In 2025, the labour force participation rate declined in the United States; Europe and Japan face even steeper demographic headwinds.

This structural shift places a natural constraint on unemployment. Even as hiring slows in sectors such as technology, unemployment rates have remained relatively stable. In many developed economies, unemployment does not rise easily because there are fewer new workers entering the labour force.

Sector experiences vary significantly. Technology lay-offs attract attention, yet technology represents a small share of total employment. At the same time, healthcare has driven roughly one-third of recent labour-force growth in the United States, reflecting ageing populations and rising care needs. In many frontline sectors, employers continue to struggle to fill roles.

When people speak about a coming wave of job destruction, I look first at demographics. In ageing societies, the deeper challenge is not too many workers. It is too few.

AI is changing tasks faster than employment

AI is already embedded in daily work. Jobseekers use it to refine resumes and applications. Employers use it to automate repetitive tasks and improve productivity. This has fuelled predictions of rapid, large-scale job loss.

The evidence does not support that outcome. Research from the Indeed Hiring Lab finds that no job today can be fully replaced by current Generative AI. Instead, only about one-quarter of jobs are likely to be highly transformed by it, rather than eliminated.

Even in occupations with high exposure to AI, such as software engineering, hiring has slowed but not stopped. AI can accelerate coding and analysis, but people remain essential for judgement, system design and accountability. At the same time, new roles are emerging in AI oversight, safety and implementation.

Adoption also remains uneven. Only about 6% of job postings mention AI, and most employers are not yet hiring specifically for AI-related roles.

Technology evolves quickly. Labour markets adjust more gradually. We have seen this before. When the internet transformed business, the long-term impact was profound, but the labour market adjusted over years, not quarters. AI is likely to follow a similar path. The transition may ultimately be large, but it will not be instantaneous.

Predictions of double-digit unemployment assume a speed of change that labour markets have rarely absorbed without severe economic disruption. If unemployment were to rise dramatically, economic output would fall sharply, and policy responses would intervene long before such conditions became permanent.

The real risk is misallocation of talent

If AI is not eliminating jobs at scale, and demographics are constraining labour supply, then the central challenge becomes clearer. It is not job scarcity. It is a misallocation.

Ageing societies cannot afford to waste talent. Yet labour markets remain slow at moving people into the roles where demand is strongest. Healthcare systems struggle to hire. Skilled trades face shortages. At the same time, some white-collar roles are being redesigned as AI absorbs routine tasks.

The question is not whether work will change. It is whether workers can move with that change.

Across many economies, time-to-hire remains elevated compared with pre-pandemic levels. Skills mismatches persist even in tight labour markets. Education systems and immigration frameworks often move more slowly than economic need. These frictions matter more in a world of constrained labour supply. When populations are shrinking, every mismatch becomes more costly.

AI can either widen or narrow this gap. Used narrowly as a cost-cutting tool, AI may reduce headcount in the short term without addressing structural shortages elsewhere. That approach risks weakening long-term resilience. In ageing economies, eliminating roles without enabling transition simply shifts pressure to other parts of the system.

Used thoughtfully, AI can help redesign jobs, redeploy workers into higher-value tasks and identify transferable skills rather than rigid credentials. It can improve matching and support mobility across sectors. The difference lies not in the technology itself, but in how leaders choose to apply it.

AI is a job transformer

Work will continue to evolve. Some tasks will disappear. Many will change. New roles will be created. That has been true in every technological era. What is distinctive today is the convergence of rapid innovation and demographic decline. In many advanced economies, the constraint is not too many workers. It is too few.

From a global, real-time view of hiring activity, one conclusion stands out: The future of work will not be defined by mass unemployment. It will be defined by how effectively we manage transition in an environment of scarce talent.

AI is not a job killer. It is a job transformer. The real risk is not automation itself, but complacency in preparing institutions, companies and workers to adapt. In ageing societies, resilience will depend less on resisting change and more on guiding it deliberately. Technology will increasingly be expected to augment people, redeploy skills and unlock productivity from talent that is becoming ever more precious.