Artificial intelligence is poised to transform economies around the world, but its impact will be deeply uneven, the head of the IMF said this week. AI is set to replace humans in an estimated 40% of jobs within the next 10 years, exacerbating inequality and requiring new social safety nets for displaced workers.
IMF Chief Sounds Alarm on AI Job Disruption
In a speech at Stanford University on Sunday, IMF Managing Director Kristalina Georgieva warned that AI and automation could displace a massive portion of the global workforce within a decade. She cited an IMF study estimating that AI could impact close to 40% of jobs worldwide, with advanced economies seeing around 60% of existing jobs transformed or eliminated.
“Tasks rather than jobs are most at risk. The good news is that new jobs will emerge across sectors,” Georgieva said. However, lower-skilled workers are likely to be hit hardest, as repetitive routine tasks get automated by AI and robotics. This could worsen income inequality and lead to social unrest, she warned.
To mitigate the disruption, Georgieva advocated for major public investment intraining programs, portable benefits for gig economy workers, and a possible universal basic income scheme funded by taxing data use. “Policymakers have a major role to play,” she emphasized, by both preparing workers for the AI economy and ensuring the technology’s benefits are broadly shared.
Advanced Economies Face Greatest Risk of Job Loss
The IMF analysis focused on advanced G20 economies, where both AI adoption and wage polarization are happening most rapidly. It estimated that across developed countries, an average of 43% of workplace tasks could soon be automated, equating to 57% of jobs facing high disruption.
Low-skilled workers in agriculture, manufacturing and utilities face the greatest threat, with over 70% of their roles at risk of full automation based on current tech capabilities. However even traditionally “safe” white collar jobs are vulnerable, with 59% of finance industry tasks and 54% in insurance now automatable. Healthcare and education remain relatively insulated for now, but are still subject to partial automation of up to 30% of occupational duties.
| Industry | Automatable Task Share |
|---|---|
| Agriculture, forestry & fishing | 71% |
| Mining & quarrying | 69% |
| Manufacturing & utilities | 71% |
| Construction | 66% |
| Wholesale, retail & hospitality | 65% |
| Transportation & storage | 64% |
| Public administration | 63% |
| Financial & insurance | 59% |
| Professional services | 44% |
| Health & social work | 23% |
| Education | 26% |
Task automability by industry. Source: IMF
Across all sectors in advanced economies, lower-skilled roles are over 50% more susceptible to being fully replaced by machines. Workers lacking tertiary education or specialized skills training are most vulnerable, the IMF concluded.
Winners and Losers in the AI Economy
The transformative power of AI will boost productivity and efficiency enormously, the IMF chief said, potentially adding $13 trillion to developed economies by 2030. However these gains may accrue more to high-skilled workers and owners of technology and data, rather than to typical employees.
“The benefits of AI and automation should be shared widely, not concentrated among a select few,” Georgieva urged. She advocated for policymakers to focus on programs that make AI’s advantages more inclusive and prepare vulnerable workers for the future.
Her calls come amidst growing alarm over AI exacerbating inequality globally. Elon Musk warned just last week that uncontrolled AI could create an “extremely unemployed future population that is very angry.” The multi-billionaire tech magnate advised governments “to put subsidies and incentives towards alleviating the AI-based employment disruption that’s coming.”
Preparing the Workforce with Reskilling and Retraining
Economists emphasize that automation anxiety tends to peak during times of technological change, before new human-AI collaboration models become established. “The debate seems to go through regular cycles of anxiety and denial,” remarked MIT scholar David Autor. While machines may take specific occupations, they can also generate demand for new specialized roles and free workers to be more creative.
Nonetheless, smooth workforce transitions will require substantial public and private investment in training, say experts. A recent McKinsey study found 87% of executives planning to use AI also intend to redesign jobs, retrain staff, and provide mobility assistance. Companies leading in AI plan to fund reskilling initiatives for over half their employees affected by automation.
Policymakers are also mobilizing large-scale retraining programs, aiming to make workforces “future-fit” for the AI-powered economy. The European Commission launched a ₤93 billion upskilling fund last November, while the US Labor Department recently announced $170 million for AI education grants. Asian tigers like Singapore and South Korea have pioneered “skillsfuture” credits and AI academies.
Ensuring vulnerable demographic groups are not left behind will be critical for an inclusive AI transition. Younger digital natives tend to adapt quicker, notes labor economist Susan Lund, whereas mid-career workers may require extensive retraining. Women also face barriers entering STEM fields, where most AI jobs are concentrated.
Bracing for Further Disruption From Generative AI
Rapid advances in generative AI could accelerate workforce disruption beyond even the IMF’s projections, add experts. Systems like ChatGPT that generate original text, images and code are gaining capabilities once considered solely human. Alongside robotic process automation that mimics routine tasks, generative models could replace administrative, creative and even technical roles.
“AI will become an interface to almost unlimited labor across many domains,” wrote AI scholar Anthropic’s Dario Amodei. As language models continue improving, more occupations could come under pressure, including researchers, writers, paralegals, accountants and radiologists.
Nonetheless, former Google CEO Eric Schmidt contends net job creation remains likely long-term, comparing AI to past innovations like electrification. While machines take rote jobs, new opportunities emerge around utilizing technology. MIT’s Autor sees openings for managers, explainers and sustainers roles assisting, monitoring and enhancing generative systems.
Policy Ideas to Smooth the Transition
Bracing for transformative economic shifts from AI, policymakers have introduced various proposals to update worker protections. As automation rips through traditional careers, labor advocates are calling for universal basic income schemes to support displaced employees. Continuing education and skills training programs have bipartisan support, aiming to reskill vulnerable cohorts.
With tech platforms and robots possibly claiming an outsized share of productivity gains, wealth taxes and digital corporate levies have sparked debate globally. IMF head Georgieva suggests taxing data flows specifically, given AI and automation’s reliance on continuously improving datasets.
Legal experts have also raised thorny issues around accountability and workplace rights regarding AI systems. If an algorithm makes hiring, firing or promotion decisions unfairly, who is liable? Updating anti-discrimination, minimum wage and occupational safety laws for an AI-hybrid world remains a complex challenge for policymakers in the 2020s.
Ensuring American workers remain competitive globally, President Harris appointed a National AI Taskforce last year. She commended it this week for issuing practical recommendations, including AI safety guidelines for firms, community college partnerships, and not restricting automation itself which could disadvantage US industry.
“AI will transform our economy and society,” Harris reiterated in a statement. “We must shape that change for the benefit of humanity.”
To err is human, but AI does it too. Whilst factual data is used in the production of these articles, the content is written entirely by AI. Double check any facts you intend to rely on with another source.
