# Jobs AI can’t replace? What the research says is least exposed > Research measures exposure to AI, not immunity. See what BLS, ILO, IMF and Pew say about the least and most exposed work, and how to judge your own role. - Canonical URL: https://jobbie.bot/blog/jobs-ai-cant-replace - Topic: [Career advice](https://jobbie.bot/guides/career-advice) - Published by [Jobbie](https://jobbie.bot/) · Updated 2026-10-06 · 8 min read Research does not identify jobs that AI cannot replace. It measures exposure: how much of a job’s work current AI could do. By that measure, the least exposed jobs are built on physical work, equipment repair, and caring for or coaching people, and the most exposed are clerical and information-processing jobs. The US Bureau of Labor Statistics (BLS) projects the fastest growth from 2025 to 2035 in healthcare occupations and a decline in office and administrative support. The sources share a caution: exposure is not the same as job loss, and the size and timing of any change are uncertain. ## Key takeaways - Studies measure exposure to AI, task by task. None of those cited here certifies a job as safe. - Least exposed in Pew Research Center’s 2023 analysis: work that depends on physical activity, repairing equipment, and assisting, caring for or coaching people. - Most exposed in the International Labour Organization’s 2025 index: clerical occupations. - BLS projects healthcare support employment to grow 13.3% from 2025 to 2035 and office and administrative support employment to fall 4.0%. - Exposed often means changed, not eliminated. The ILO calls transformation of jobs the most likely effect. ## What “exposed to AI” means, and what it does not An exposure study breaks an occupation into tasks or work activities, judges how much of each one AI could do, and adds up the result. That shows where the technology overlaps with the work. It does not show what employers will do about it. Pew writes that because AI could be used either to replace or to complement what workers do, “it is not known exactly which or how many jobs are in peril.” The International Monetary Fund (IMF) estimates that almost 40% of global employment is exposed to AI, rising to about 60% of jobs in advanced economies, and that roughly half of those exposed jobs may benefit through higher productivity. For the other half, it says AI could take over key tasks, which could lower demand for labor, wages and hiring. The sources offer two kinds of evidence: exposure studies, which rate what AI could do, and employment projections, which estimate how many people each occupation will employ. ## Which jobs are least exposed to AI? Pew rated 41 work activities across 873 occupations, using O\*NET, the US Department of Labor’s occupational database. It judged nine of the activities to have low exposure. The ones it names include assisting and caring for others, performing general physical activities, repairing equipment, and coaching and developing others. Its examples of less exposed jobs are barbers, child care workers and nannies. In Pew’s count, 23% of US workers were in the least exposed jobs. Pew notes that its analysis “doesn’t consider the role of AI-enabled machines or robots that may perform mechanical or physical tasks.” Projections add a second view, with AI as only one force among several. For 2025–35, BLS expects: - **Healthcare to grow fastest.** Of the 22 major occupational groups, healthcare support (13.3%) and healthcare practitioners and technical occupations (8.0%) are projected to grow the most, against 3.5% for all occupations. BLS points to an aging population and the prevalence of chronic conditions. Nurse practitioners lead all occupations at 41.0%. - **Two hands-on energy occupations to follow.** Solar photovoltaic installers (36.5%) and wind turbine service technicians (29.5%) are projected to grow on demand for installing solar panels and maintaining turbines. Growth is not immunity. These projections reflect rising demand for the work, not a finding that it is out of AI’s reach. ## Which jobs are most exposed to AI? The clearest agreement across the sources is on clerical and routine information work. - **Clerical occupations.** The ILO’s 2025 index finds they have the highest exposure to generative AI of any group. It also finds rising exposure in some strongly digitized occupations, which it links to AI’s growing ability with specialized tasks in professional and technical roles. - **Office and administrative support.** BLS projects this group to decline 4.0% from 2025 to 2035 and lose 752,100 jobs, the most of any major group. It says automation tools, including those powered by AI, are likely to reduce demand for several of these occupations. - **Arts, design and media, and sales.** BLS says generative AI may limit demand for some jobs in arts, design, entertainment, sports and media, and expects e-commerce and AI sales tools to contribute to a 1.4% decline in sales occupations. - **Specific occupations in the 2023–33 projections.** BLS projected declines for customer service representatives (5.0%) and medical transcriptionists (4.7%), whose core tasks it said generative AI could most easily replicate, and for auto damage insurance appraisers (9.2%), claims adjusters, examiners and investigators (4.4%) and credit analysts (3.9%). Pew’s high-exposure examples are budget analysts, data entry keyers, judicial law clerks and web developers. Among the activities it rates as highly exposed are getting information and analyzing data. Globally, the ILO finds one in four workers in an occupation with some exposure and 3.3% of employment in its highest category, with more exposure in high-income countries (34%) than in low-income ones (11%). ## Why exposed does not mean replaced Several occupations that AI touches are still projected to grow. | Occupation | What BLS says about AI | Projected change | | --- | --- | --- | | Software developers | AI can assist with coding, but software demand is expected to stay strong | Up 17.9%, 2023–33 | | Lawyers | Accuracy matters in legal settings, so AI output needs human review | Up 5.2%, 2023–33 | | Paralegals and legal assistants | The strongest productivity effect among the legal occupations studied | Up 1.2%, 2023–33 | | Data scientists | In demand to develop and apply these technologies | Up 34.6%, 2025–35 | The legal pair shows the pattern: inside one field, the role built on document review is projected to grow far more slowly than the role that carries the accountability. The ILO reaches a similar conclusion at global scale. Because most occupations consist of tasks that require human input, it expects transformation of jobs to be the most likely effect of generative AI. ## How certain are these findings? Less certain than a list of “AI-proof jobs” suggests. - **BLS says its methods have limits.** They “are not designed to capture extremely rapid technological change,” and “the timing and scale of many potential impacts of GenAI are too uncertain to be reflected in BLS projections.” - **Exposure indexes are judgments about capability.** They rate what AI could do with a task. They do not count jobs that have changed. - **The findings date quickly.** The ILO revised its 2023 index in 2025 and found exposure had risen for some occupations. Pew’s analysis is from 2023 and leaves out robots. - **The net effect is open.** The IMF calls it difficult to foresee. Check the date and method of anything you read on this subject. ## How to judge your own role’s exposure The research works task by task, and so can you. 1. **List what you do in a typical week.** Write ten to fifteen tasks, each starting with a verb. 2. **Sort each task by type.** Use the categories from the research: getting and analyzing information (rated more exposed), physical work and repair (less exposed), assisting, caring for or coaching people (less exposed), and work where someone must review and answer for the result. 3. **Estimate the hours in each group.** The split matters more than the job title. 4. **Test the exposed tasks.** Try a current AI tool on real examples and check the output, the same habit that applies when you [use AI in your job search](https://jobbie.bot/blog/how-to-use-ai-in-your-job-search). 5. **Shift your time on purpose.** Learn the tools for the exposed tasks, ask for more of the other work, and write both into a [career development plan](https://jobbie.bot/blog/career-development-plan). ```Task audit: one entry per task Task: [what you do, starting with a verb] Hours per week: [number] Type: [information / physical / people / review and sign-off] Can a current AI tool do most of it? [yes / partly / no] How I know: [tried it / seen it used / guessing] What still needs me: [the part a tool cannot do or answer for] ``` If most of your week lands in the exposed group, start from your [transferable skills](https://jobbie.bot/blog/transferable-skills). If you decide to move, follow a deliberate [career change plan](https://jobbie.bot/blog/career-change) instead of chasing a list of safe jobs. ## Frequently asked questions ### Which jobs are least exposed to AI? In Pew Research Center’s 2023 analysis, the least exposed jobs are those whose most important activities are physical work, repairing equipment, and assisting, caring for or coaching people. Pew did not assess robots. Separately, BLS projects healthcare occupations to grow fastest from 2025 to 2035. Neither source calls any job safe. ### Which jobs are most exposed to AI? The International Labour Organization’s 2025 index finds that clerical occupations have the highest exposure to generative AI. BLS projects US office and administrative support employment to fall 4.0% from 2025 to 2035, the largest loss of any major occupational group, and says automation tools including AI are likely to reduce demand for several of those occupations. ### Does high exposure to AI mean a job will disappear? No. Exposure measures how much of the work AI could do, not what will happen to the job. The ILO expects transformation of jobs to be the most likely effect, because most occupations include tasks that need human input. The IMF estimates that roughly half of exposed jobs in advanced economies may benefit from AI. ### Are skilled trades safe from AI? None of the research cited here describes any job as safe. Pew rated physical activities and equipment repair as low exposure in 2023, but its analysis did not consider AI-enabled machines or robots. BLS projects fast growth to 2035 for solar photovoltaic installers and wind turbine service technicians, which it attributes to demand for that work. ### Should I change careers because of AI? Not on the strength of a list. Look at your own tasks first: which are the information tasks that research rates as more exposed, and which depend on physical presence, care for people or accountable judgment. Move your time and skills toward the second group and learn the tools for the first. ## Sources - [BLS: Employment Projections, 2025–2035 summary](https://www.bls.gov/news.release/archives/ecopro_08272026.htm): projected change by occupational group, the fastest-growing occupations and where BLS expects AI to affect demand (August 2026). - [BLS Monthly Labor Review: Incorporating AI impacts in BLS employment projections](https://www.bls.gov/opub/mlr/2025/article/incorporating-ai-impacts-in-bls-employment-projections.htm): the 2023–33 occupational case studies and BLS’s statements on uncertainty (February 2025). - [ILO: Generative AI and Jobs: A Refined Global Index of Occupational Exposure](https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure): global exposure shares, the most exposed occupations and the expected transformation of jobs (May 2025). - [IMF: AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity.](https://www.imf.org/en/Blogs/Articles/2024/01/14/ai-will-transform-the-global-economy-lets-make-sure-it-benefits-humanity): exposure by country group and how many exposed jobs may benefit (January 2024). - [Pew Research Center: Which U.S. Workers Are More Exposed to AI on Their Jobs?](https://www.pewresearch.org/social-trends/2023/07/26/which-u-s-workers-are-more-exposed-to-ai-on-their-jobs/): the share of US workers in the most and least exposed jobs and the activities behind the ratings (July 2023).