Daniel Makina and Rogers Dhliwayo
The rapid diffusion of artificial intelligence (AI) across global economies is transforming the nature of work with profound implications for labour markets, skills development, and social equity. Kristalina Georgieva, the Managing Director of the International Monetary Fund (IMF) highlights that although AI and digital technologies have historically reshaped employment, today’s transformation is distinctive in both scale and scope, with nearly 40 percent of jobs globally exposed to AI-driven change and new skills increasingly central to economic participation.
Yet the transformative potential of AI is unevenly distributed across the global economy. Reporting for the Financial Times, analysts cite warnings from Anthropic’s head of economics, Peter McCrory, that faster AI adoption in advanced economies risks widening global inequality if productivity gains remain concentrated in high-income countries. This concern is reinforced by evidence of rapid innovation concentration at the technological frontier.
The IMF’s AI Preparedness Index shows that high-income countries are substantially better positioned to capture AI-driven gains, while many low-income economies continue to face binding constraints in basic electricity, broadband, and foundational digital infrastructure. Taken together, these dynamics highlight a growing tension between technological acceleration and inclusive development that future global policy frameworks must confront directly.
The diffusion of AI is fundamentally reshaping skills demand across labour markets. High-growth occupations increasingly prioritize digital literacy, data analysis, and human–AI collaboration, while routine and many middle-skill roles are being reconfigured or displaced. IMF evidence indicates that vacancies requiring new skills – particularly in information technology are associated with wage premiums; however, these gains have not translated into net employment growth in AI-exposed occupations, resulting in weaker overall employment outcomes in regions with high concentrations of AI-intensive work. The result is an increasingly polarized labour market in which workers possessing AI-complementary skills capture disproportionate benefits relative to those whose skills are more easily substituted.
Micro-level evidence reinforces this pattern; it has been shown that demand for AI-complementary capabilities, such as digital proficiency, problem-solving, and teamwork have expanded more rapidly than demand for automation-susceptible skills. This shift is accompanied by rising wage premiums across both AI-intensive and non-AI roles, underscoring that complementarities rather than automation alone are driving labour market rewards. Nevertheless, researchers caution that substitution effects remain significant for specific tasks, highlighting the continued need for targeted retraining and reskilling interventions to mitigate displacement risks.
Complementing this perspective, Drake Mullens and Stella Shen conceptualize labour market adjustment through their “AI-Accentuated Career Transitions” (2ACT) framework, which emphasizes how AI adoption interacts with existing skill portfolios to shape occupational mobility. Their findings suggest that workers who combine cognitive and technical skills with iterative, task-embedded AI use are most likely to experience upward career transitions. This reinforces the importance of sustained skill accumulation and continuous learning, rather than one-off training responses, in navigating AI-driven labour market change.
The international distribution of these dynamics further complicates the picture. Analysis reported by the Financial Times draws on Anthropic research to show that AI is deployed more extensively in workplace settings in high-income economies such as the United States, the United Kingdom, and Japan, amplifying productivity effects, while lower-income countries tend to concentrate AI use in educational or experimental contexts with more limited short-term economic impact.
Peter McCrory warns that absent deliberate efforts to broaden AI literacy and sectoral adoption, productivity gains may increasingly diverge across countries, reinforcing global income disparities. These concerns echo warnings from multilateral institutions, including the World Trade Organization (WTO) and UNDP that AI could entrench rather than alleviate inequality if diffusion remains uneven.
Broader academic research corroborates these distributional risks; it finds that automation-oriented AI adoption tends to depress employment and wages in low-skill occupations, whereas augmentation-oriented applications generate new roles and wage growth for high-skill workers, contributing to widening within-country wage inequality. At the same time, research documents a growing shift toward skill-based hiring in AI- and green-related occupations, with employers placing less emphasis on formal degree credentials. This suggests that more flexible skill pathways can, under the right conditions, expand access to high-growth opportunities.
For South Africa, these dynamics are particularly consequential. The labour market is already characterized by persistent structural unemployment – most acute among youth – and deep skills mismatches. In this context, limited expansion of AI-relevant skills risks compounding existing exclusion. As the IMF staff discussion note emphasizes that closing future skill gaps requires not only formal education reform but also adaptive lifelong learning systems capable of responding to rapidly evolving labour demand. Without such institutional capacity, South Africa risks reproducing the polarized outcomes observed in advanced economies, where the gains from AI accrue to a narrow segment of skilled workers while the majority face stagnation or displacement.
Policy responses in South Africa should therefore be multi-pronged. First, integration of AI-oriented competencies into education and training curricula across levels is essential. This includes strengthening science, technology, engineering, and mathematics (STEM) education, digital literacy, and problem-solving skills, while also embedding socio-cognitive competencies such as ethical reasoning and teamwork that complement AI technologies. Research suggests that human-centric skills remain resilient even as technical competencies shift, and countries that cultivate both are better positioned to harness AI for inclusive growth.
Second, lifelong learning and upskilling infrastructures must be expanded. This goes beyond formal qualifications to include industry-recognized certifications, modular online learning, and employer-led training partnerships. The IMF’s Skill Imbalance Index findings suggest that countries with stronger training ecosystems can better align skill supply with dynamic labour demand. South Africa can leverage public–private collaborations, including partnerships with technology firms and education platforms, to bolster access to AI-related training at scale.
Third, labour market policies should be recalibrated to provide safety nets and active support for displaced workers. This includes unemployment benefits linked to reskilling participation, targeted job matching services, and incentives for firms that invest in internal training. Without such structures, transitions to new occupations can be slow and socially costly.
Fourth, innovation and regulatory policy must be aligned with labour market goals. South Africa should nurture AI innovation ecosystems that support local research, entrepreneurship, and technology adaptation suited to the country’s development needs. Regulatory frameworks should balance competitiveness with worker protection, ensuring ethical AI deployment and guarding against exploitative practices.
Fifth, cross-sector coordination is essential. Alignment between ministries of education, labour, and economic development can help bridge institutional silos that often hamper systemic skill transformation. By fostering a shared strategic vision, one that integrates AI readiness into national development plans – South Africa can avoid the “middle skill squeeze” and ensure that the economic dividends from AI are distributed broadly across society.
In conclusion, AI’s role as a labour market disruptor and enhancer presents both risk and opportunity. For South Africa, the challenge lies not in resisting technological change, but in shaping policies that align skill development with inclusive economic participation. By investing in education, fostering lifelong learning, and strengthening labour market institutions, South Africa can transform the AI transition from a source of disruption into a catalyst for broad-based prosperity.
Daniel Makina and Rogers Dhliwayo are editors of Economic Business Insights
