ARTIFICIAL INTELLIGENCE

“We assess that 40% of jobs globally are going to be impacted by AI over the next couple of years – either enhanced, eliminated, or transformed. In advanced economies, it’s 60%.” – Kristalina Georgieva – Managing Director, IMF

Kristalina Georgieva’s assessment of AI’s labour market impact represents one of the most consequential economic forecasts of our time. Speaking at the World Economic Forum in Davos in January 2026, the Managing Director of the International Monetary Fund articulated a sobering reality: artificial intelligence is not a distant threat but an immediate force already reshaping employment globally. Her invocation of a “tsunami”-a natural disaster of overwhelming force and scale-captures the simultaneity and inevitability of this transformation.

The Scale of Disruption

Georgieva’s figures warrant careful examination. The IMF calculates that 40 per cent of jobs globally will be touched by AI, with each affected role falling into one of three categories: enhancement (where AI augments human capability), elimination (where automation replaces human labour), or transformation (where roles are fundamentally altered without necessarily improving compensation). This is not speculative projection but empirical assessment grounded in IMF research across member economies.

The geographical disparity is striking and consequential. In advanced economies-the United States, Western Europe, Japan, and similar developed nations-the figure reaches 60 per cent. By contrast, in low-income countries, the impact ranges from 20 to 26 per cent. This divergence is not accidental; it reflects the concentration of AI infrastructure, capital investment, and digital integration in wealthy nations. The IMF’s concern, as Georgieva articulated, is what she termed an “accordion of opportunities”-a compression and expansion of economic possibility that varies dramatically by geography and development status.

Understanding the Context: AI as Economic Transformation

Georgieva’s warning must be situated within the broader economic moment of early 2026. The global economy faces simultaneous pressures: geopolitical fragmentation, demographic shifts, climate transition, and technological disruption occurring in parallel. AI is not the sole driver of economic uncertainty, but it is perhaps the most visible and immediate.

The IMF’s analysis distinguishes between AI’s productivity benefits and its labour market risks. Georgieva acknowledged that AI is generating genuine economic gains across sectors-agriculture, healthcare, education, and transport have all experienced productivity enhancements. Translation and interpretation services have been enhanced rather than eliminated; research analysts have found their work augmented by AI tools. Yet these gains are unevenly distributed, and the labour market adjustment required is unprecedented in speed and scale.

The productivity question is central to Georgieva’s economic outlook. Global growth has been underwhelming in recent years, with productivity growth stagnant except in the United States. AI represents the most potent force for reversing this trend, with potential to boost global growth between 0.1 and 0.8 per cent annually. A 0.8 per cent productivity gain would restore growth to pre-pandemic levels. Yet this upside scenario depends entirely on successful labour market adjustment and equitable distribution of AI’s benefits.

The Theoretical Foundations: Labour Economics and Technological Disruption

Georgieva’s analysis draws on decades of labour economics scholarship examining technological displacement. The intellectual lineage traces to economists such as David Autor, who has extensively studied how technological change reshapes labour markets. Autor’s research demonstrates that whilst technology eliminates routine tasks, it simultaneously creates demand for new skills and complementary labour. However, this adjustment is neither automatic nor painless; workers displaced from routine cognitive tasks often face years of unemployment or underemployment before transitioning to new roles.

The “task-based” framework of labour economics-developed by scholars including Autor and Frank Levy-provides the theoretical scaffolding for understanding AI’s impact. Rather than viewing jobs as monolithic units, this approach recognises that occupations comprise multiple tasks. AI may automate certain tasks within a role whilst leaving others intact, fundamentally altering job content and skill requirements. A radiologist’s role, for instance, may be transformed by AI’s superior pattern recognition in image analysis, but the radiologist’s diagnostic judgment, patient communication, and clinical decision-making remain valuable.

Erik Brynjolfsson and Andrew McAfee, prominent technology economists, have argued that AI represents a qualitative shift from previous technological waves. Unlike earlier automation, which primarily affected routine manual labour, AI threatens cognitive work across income levels. Their research suggests that without deliberate policy intervention, AI could exacerbate inequality rather than reduce it, concentrating gains among capital owners and highly skilled workers whilst displacing middle-skill employment.

Daron Acemoglu, the MIT economist, has been particularly critical of “so-so automation”-technology that increases productivity marginally whilst displacing workers without creating sufficient new opportunities. His work emphasises that technological outcomes are not predetermined; they depend on institutional choices, investment priorities, and policy frameworks. This perspective is crucial for understanding Georgieva’s policy recommendations.

The Policy Imperative

Georgieva’s framing of the challenge as a policy problem rather than an inevitable outcome reflects this economic thinking. She has consistently advocated for three policy pillars: investment in skills development, meaningful regulation and ethical frameworks, and ensuring AI’s benefits penetrate across sectors and geographies rather than concentrating in advanced economies.

The IMF’s own research indicates that one in ten jobs in advanced economies already require substantially new skills-a figure that will accelerate. Yet educational and training systems globally remain poorly aligned with AI-era skill demands. Georgieva has urged governments to invest in reskilling programmes, particularly targeting workers in roles most vulnerable to displacement.

Her emphasis on regulation and ethics reflects growing recognition that AI’s trajectory is not technologically determined. The choice between AI as a tool for broad-based productivity enhancement versus a mechanism for labour displacement and inequality concentration remains open. This aligns with the work of scholars such as Shoshana Zuboff, who argues that technological systems embody political choices about power distribution and social organisation.

The Global Inequality Dimension

Perhaps most significant is Georgieva’s concern about the “accordion of opportunities.” The 60 per cent figure for advanced economies versus 20-26 per cent for low-income countries reflects not merely different levels of AI adoption but fundamentally different economic trajectories. Advanced economies possess the infrastructure, capital, and institutional capacity to invest in AI whilst simultaneously managing labour market transition. Low-income countries risk being left behind-neither benefiting from AI’s productivity gains nor receiving the investment in skills and social protection that might cushion displacement.

This concern echoes the work of development economists such as Dani Rodrik, who has documented how technological change can bypass developing economies entirely, leaving them trapped in low-productivity sectors. If AI concentrates in advanced economies and wealthy sectors, developing nations may face a new form of technological colonialism-dependent on imported AI solutions without developing indigenous capacity or capturing value creation.

The Measurement Challenge

Georgieva’s 40 per cent figure, whilst grounded in IMF research, represents a probabilistic assessment rather than a precise prediction. The IMF acknowledges a “fairly big range” of potential impacts on global growth (0.1 to 0.8 per cent), reflecting genuine uncertainty about AI’s trajectory. This uncertainty itself is significant; it suggests that outcomes remain contingent on policy choices, investment decisions, and institutional responses.

The distinction between jobs “touched” by AI and jobs eliminated is crucial. Enhancement and transformation may be preferable to elimination, but they still require worker adjustment, skill development, and potentially geographic mobility. A job that is transformed but offers no wage improvement-as Georgieva noted-may be economically worse for the worker even if technically retained.

The Broader Economic Context

Georgieva’s warning arrives amid broader economic fragmentation. Trade tensions, geopolitical competition, and the shift from a rules-based global economic order toward competing blocs create additional uncertainty. AI development is increasingly intertwined with strategic competition between major powers, particularly between the United States and China. This geopolitical dimension means that AI’s labour market impact cannot be separated from questions of technological sovereignty, supply chain resilience, and economic security.

The IMF chief has also emphasised that AI’s benefits are not automatic. She personally undertook training in AI productivity tools, including Microsoft Copilot, and urged IMF staff to embrace AI-based enhancements. Yet this individual adoption, multiplied across millions of workers and organisations, requires deliberate choice, investment in training, and organisational restructuring. The productivity gains Georgieva projects depend on this active embrace rather than passive exposure to AI technology.

Implications for Policy and Strategy

Georgieva’s analysis suggests several imperatives for policymakers. First, labour market adjustment cannot be left to market forces alone; deliberate investment in education, training, and social protection is essential. Second, the distribution of AI’s benefits matters as much as aggregate productivity gains; without attention to equity, AI could deepen inequality within and between nations. Third, regulation and ethical frameworks must be established proactively rather than reactively, shaping AI development toward socially beneficial outcomes.

Her invocation of a “tsunami” is not mere rhetoric but a precise characterisation of the challenge’s scale and urgency. Tsunamis cannot be prevented, but their impact can be mitigated through preparation, early warning systems, and coordinated response. Similarly, AI’s labour market impact is largely inevitable, but its consequences-whether broadly shared prosperity or concentrated disruption-remain subject to human choice and institutional design.