AI and robots fit the industries that are ageing least, BIS finds

Between 2010 and 2024, the share of agricultural workers aged 55 and over rose from 24% to 33% worldwide. In health and social work it climbed from 18% to 25%. A BIS Bulletin published on 24 September 2026 finds that these two ageing industries sit near the bottom of the scale for automation by AI and robots, while the industries those technologies fit best have younger staff.

Where the machines fit, and where the older workers are

[Fact] Iñaki Aldasoro, Sebastian Doerr and Daniel Rees of the Bank for International Settlements combined ILO employment data by age and industry for 135 economies with two existing exposure measures: the AI industry exposure index of Felten, Raj and Seamans (2021) and robot density data from Acemoglu and Restrepo (2020). They averaged the two into one score per industry.

[Fact] Manufacturing ranks first on that score (1.91 standard deviations above the industry average), followed by finance and insurance (0.91), professional and technical services (0.72), education (0.58) and information and communication (0.43). Manufacturing owes its rank to robots alone: on AI exposure it scores -0.42, below the average. Finance has the highest AI exposure of any industry, at 2.08. Agriculture is last at -0.94, after construction (-0.69) and accommodation and food (-0.65).

[Fact] Health and social work scores 0.19. The authors describe it as large, ageing and only weakly exposed, with care and interpersonal tasks that remain hard to automate, and with demand that grows as populations age. Agriculture employs over 700 million people, and its median share of national employment across the sample is 14.2%.

A country score: the ageing-automation overlap

[Fact] For each economy, the authors measure whether older workers are concentrated in the industries that AI and robots can reach, weighted by how many people each industry employs. A higher score means ageing falls in automatable industries. The score is highest in North America, parts of northern and western Europe and Australia. It is low across much of southern and Southeast Asia and in parts of the Balkans and Caucasus.

[Fact] The link with income is weak. The correlation between the score and log GDP per capita is 0.14, rising to 0.36 when sub-Saharan Africa is left out. Several fast-ageing advanced economies, including Asian ones, have among the worst scores, because their automation capacity sits in a large manufacturing sector with a young workforce. The authors conclude that leading the world in AI or robot adoption does not mean an economy is well placed to automate an ageing workforce.

[Fact] The fit has worsened. Compared with 2010-2013, the score fell in more than half of the economies, with a median change of -0.09. For East and Southeast Asia and for southern Europe, projected ageing to 2050 is fast while the score is low.

[Fact] Where the score is low, the authors point to higher labour force participation, labour mobility and immigration. Economies with low scores currently have smaller migrant populations, and the score is almost uncorrelated with a measure of labour market flexibility.

What workers in the ageing industries can prepare

The bulletin works with industries, not occupations. [Estimate] The jobs most directly inside its slowest-to-automate, fastest-ageing groups include home health aides, nursing assistants, registered nurses and crop and greenhouse farmworkers.

• Care workers: [Estimate] Rising demand, a low exposure score and an older workforce point to continued hiring pressure in health and social work. Credentials that let you move from aide to nursing or supervisory roles hold value in an industry where retirements are not being filled by machines.
• Farm workers: [Claim] The authors note that agriculture in some economies may still have scope to mechanise with existing technology. Training on machinery already in use is the change most likely to reach farm jobs first.
• Younger workers in finance and professional services: [Fact] These industries have the highest AI exposure and comparatively young staff. [Estimate] AI adoption there will not be driven by retirements, so financial and investment analysts and similar staff will see their tasks change mid-career.
• Anyone over 50 considering a switch: [Fact] The authors call labour mobility critical for helping workers change occupations, and cite OECD evidence that job mobility declines with age. [Estimate] A move made in your early 50s is likely easier than one made near retirement.

Limits

Exposure here means technical feasibility, not whether AI or robots replace or assist workers, and the measure is static: it does not allow for new technology, retraining or workers changing industries. The AI scores are aggregated with US employment weights and applied to every economy, Saudi Arabia is missing from the industry data, and China's figures come from its 2020 census; the per-country table in the online annex is an image and was not read for this summary.

Sources

• Iñaki Aldasoro, Sebastian Doerr and Daniel Rees, "Old workers, young machines: can AI and automation offset population ageing?", BIS Bulletin No 135, 24 September 2026: https://www.bis.org/publications/bulletin-135-old-workers-young-machines-can-ai-and-automation-offset-population-ageing

About this analysis

This article was written with AI assistance, from the full text of the bulletin and its online annex. Items marked [Estimate] (the occupation mapping and the advice drawn from industry-level results) are our own reading of the bulletin. The authors state that they also used AI for research assistance and editorial refinement. No graphs or tables from the bulletin are reproduced.