Digital Collar
Menu

AI and older workers: what the research says and what helps

Updated

Discussions of AI at work often assume older workers will be left behind. The research is more careful than that. It does point to real risks, mainly around training and access to new skills, but it also warns that some of what is said about older workers and technology rests on assumptions rather than evidence. This page sets out what two primary sources say and what helps.

The short answer: the OECD finds that older workers can be more vulnerable when AI is introduced at work, mainly because they are on average less likely to take part in training, and it recommends that AI training reach them specifically. It also notes that beliefs about older workers being skeptical of AI come from colleagues' perceptions, which may be biased. Experience, judgment and knowledge of how work really gets done are exactly what checking and supervising AI requires, so the right training and support matter more than age.

What the OECD found

The OECD Employment Outlook 2023, drawing on employer and worker surveys in manufacturing and finance and on wider research, reports that (OECD Employment Outlook 2023):

  • Employers expected mixed effects: for older and low-skilled workers, surveyed employers were more likely to say AI would harm rather than benefit them, while almost half thought AI would help workers with disabilities.
  • Training access is the main gap: older adults are less likely to take part in adult learning in every country in the OECD's adult skills survey; the average participation gap between people aged 54 and over and those aged 25 to 53 was about 22 percentage points in 2015.
  • Some AI systems are harder to use for people with low digital skills, and the OECD notes that more than a third of adults on average lack even the most basic digital skills, across age groups.
  • Perceptions may be biased: older workers were described by colleagues as skeptical of AI, but the OECD notes the study did not report older workers themselves voicing that, and that such beliefs may reflect biases rather than actual abilities and attitudes.
  • AI can cut both ways on fairness: AI-supported evaluations could bring better recognition for groups that have faced bias, including older workers, but if AI replicates existing biases the effect could be the opposite.

Its policy conclusion is that training for AI should reach vulnerable groups, older workers and the low-skilled in particular, as well as higher-skilled workers and managers.

The wider demographic picture

The WEF's Future of Jobs Report 2025 lists aging and declining working-age populations, mainly in higher-income economies, among the trends employers expect to transform their business, and links them to growing demand for skills in talent management, teaching and mentoring (WEF Future of Jobs Report 2025). The OECD also notes that some companies said AI could help relieve skills needs in the face of an aging population and labor shortages.

Strengths that matter more with AI

As AI takes on drafting and routine processing, the work that remains leans on things that come with experience:

  • Knowing what right looks like, which is what you need to catch confident but wrong AI output. See checking AI output.
  • Judgment about exceptions that the standard process does not cover.
  • Relationships with clients and colleagues built over time.
  • Teaching and mentoring, skills the WEF links to demographic change.
  • Supervising AI systems, which is closer to managing than to coding. See human in the loop and accountability.

Practical steps for workers

  1. Ask for training on the tools your workplace uses, and take it early.
  2. Start with your own tasks: use AI on one or two things you know well, so you can judge the output. See how to use AI at work.
  3. Position your experience as quality control: volunteer to review AI output or help set the rules.
  4. Learn in order, rather than chasing every tool. See AI skills to learn.
  5. Plan ahead: see how to future-proof your career from AI.

Practical steps for employers

  • Offer training to everyone, and check who is actually taking part. See how to train employees on AI.
  • Design tools that are easy to use, with support for people who need it.
  • Do not assume attitudes by age; ask people directly.
  • Check AI used for evaluation or hiring for bias. The NIST AI RMF lists fairness with harmful bias managed among the characteristics of trustworthy AI (NIST AI RMF 1.0). See AI monitoring at work, AI in hiring and AI and disabled workers.
  • Use experienced staff as reviewers and mentors in AI rollouts.

Questions and answers

How does AI affect older workers?

The OECD finds older workers can be more vulnerable when AI is introduced, mainly because they are on average less likely to take part in training, and recommends AI training reach them specifically.

Are older workers worse at using AI?

The evidence does not show that. The OECD notes that claims of older workers being skeptical of AI came from colleagues’ perceptions, which may be biased, not from older workers themselves.

What skills help older workers with AI?

Experience-based skills such as knowing what good work looks like, judgment about exceptions, relationships and mentoring, which are central to checking and supervising AI.

Should employers train older workers on AI?

Yes. The OECD recommends that training for AI reach vulnerable groups, older workers and the low-skilled in particular, as well as higher-skilled workers and managers.

Can AI make workplaces fairer for older workers?

It can, if AI-supported evaluations reduce existing bias. If AI replicates existing biases, the OECD warns the effect could be the opposite, so systems need checking.

Sources