Most AI-resilient careers: what the evidence shows
By the Mandlys editorial team · Updated
Occupations with physical work in unpredictable settings, close human contact and real personal accountability are the least exposed to AI according to research. Examples include healthcare and care work, skilled trades and installation, and leadership roles. No job is fully AI-proof, but in these occupations AI more often changes part of the work than replaces it.
Traits that make an occupation less exposed
Task-based research (Autor, Levy and Murnane 2003; Arntz et al. 2016) and the ILO generative AI index point to the same pattern. Exposure is lower when work requires:
- Physical work in varied environments, such as electrical installation, repair and construction.
- Care and trust, such as nursing, therapy and teaching young children.
- Leadership and accountability, where someone must make and own decisions.
- Negotiation and persuasion in unexpected situations.
Occupational groups with lower exposure
The ILO analyses from 2023 and 2025 show that exposure to generative AI is lowest in groups dominated by manual work, such as craft workers, machine operators and agricultural workers, and highest in clerical work. Generative AI mainly affects knowledge and office work, whereas earlier automation hit routine physical work harder.
Within knowledge work, roles built on relationships and judgement, such as managers, teachers, psychologists and nurses, are generally less exposed than roles dominated by processing text and data.
💡 Want to see how this applies to your own job? The free test takes about five minutes and needs no account. → Take the free AI test
Why no job is fully safe
Even low-exposure occupations get new tools: documentation in healthcare, scheduling in trade businesses, admin for managers. People who learn the tools are in a stronger position. The difference is often within an occupation rather than between occupations: two people with the same title can have very different task mixes.
How to use this list
Treat the groups as a starting point, not a verdict. Look at the tasks that fill your week. Read how to assess your own risk or take the test for a structured estimate.
Frequently asked questions
Which jobs are most AI-proof?
Jobs with physical work in varied settings, close human contact and real accountability, such as healthcare, skilled trades, installation and leadership roles, are the least exposed according to the ILO and task-based research.
Are there completely AI-proof jobs?
No. Even low-exposure occupations are affected by new tools, but AI more often changes part of the work than replaces it.
Which jobs are most exposed to generative AI?
Clerical and administrative work is the most exposed group according to the ILO, because many tasks involve handling text and data.
Sources
- Gmyrek, Berg & Bescond (2023). Generative AI and Jobs: A global analysis of potential effects on job quantity and quality. ILO Working Paper 96.
- Gmyrek, Berg, Kamiński m.fl. (2025). Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO Working Paper 140.
- Autor, Levy & Murnane (2003). The Skill Content of Recent Technological Change. Quarterly Journal of Economics, 118(4).
- Arntz, Gregory & Zierahn (2016). The Risk of Automation for Jobs in OECD Countries. OECD Social, Employment and Migration Working Papers No. 189.