Will AI take our jobs?

One conclusion is consistent across major institutions: AI impact is uneven, and it is primarily task-based rather than “job-based.” In other words, AI doesn’t replace occupations in one step; it automates or accelerates specific activities inside a role. That’s why the most useful unit of analysis is the task mix within each job—not the job title itself.

The IMF quantifies how broad this exposure is. Its analysis suggests that almost 40% of global employment is exposed to AI. Exposure is significantly higher in advanced economies (about 60% of jobs) than in emerging market economies (about 40%) and low-income countries (about 26%). Critically, “exposed” does not mean “eliminated.” The IMF also stresses that AI can complement many roles: in advanced economies, roughly half of exposed jobs could be negatively affected, while the remainder may benefit through higher productivity and task augmentation.

The World Economic Forum frames the next phase less as pure job destruction and more as job reallocation and churn. In its Future of Jobs Report 2025 outlook, structural labour-market transformation is projected to affect 22% of today’s jobs by 2030, with 170 million new roles created and 92 million displaced (a net increase of 78 million). The key message is that the labour market is being reshaped through simultaneous decline and growth across different role families—not a uniform collapse of employment.

The ILO’s task-level work helps explain which tasks sit closest to the automation frontier—especially office, administrative, and text-heavy work, where generative AI is particularly strong. In its global analysis, clerical work is the only broad occupation group rated as highly exposed, with 24% of clerical tasks considered highly exposed and an additional 58% at medium exposure. For most other occupational groups, the share of highly exposed tasks is far smaller (typically 1–4%). The practical implication is that the dominant near-term effect is augmentation—automating slices of a job—more often than fully automating entire occupations.

Put together, the data supports a practical answer to “Will AI take our jobs?”: AI will take some tasks—often quickly—especially language- and document-centric tasks. But full job loss usually takes longer and depends on whether organizations redesign workflows, accountability, and role definitions around automation. In many cases, the more realistic outcome is not disappearance, but job redesign: new task bundles, higher baseline expectations, and new roles created to build, operate, and govern AI-enabled processes.