Table of Contents
- 1. Look at Growing Jobs, Not Job-Loss Headlines — WEF and BLS Data
- 2. Three Principles of AI-Resistant Work — Physical, Responsibility, Relationships
- 3. Category 1: Physical Jobs — Healthcare, Care Work and the Trades
- 4. Category 2: High-Responsibility Judgment — Clinicians, Lawyers, Management
- 5. Category 3: Creativity × Relationships — Counselors, Education, Direction
- 6. Category 4: The People Who Run the AI — Moving Opposite to Pure Coding Jobs
- 7. 15 Growing Jobs — BLS Projections, Pay and Entry Requirements
- 8. Four Moves From Your Current Career to the Growing Side
- Summary
- FAQ
You have probably seen more than enough articles about jobs AI will take. Articles about jobs AI is unlikely to take — or that are actually growing are much rarer. The World Economic Forum's Future of Jobs Report 2025 expects that from 2025 to 2030 about 92 million jobs will be displaced while about 170 million are created, for a net gain of about 78 million. If work is growing on net, why do job-loss stories dominate? Because the jobs that shrink and the jobs that grow are not the same occupations.
This article has one job: to lay out the conditions that make work likely to last, and the concrete requirements for getting into growing occupations. Which occupations are most affected is covered in 15 Jobs Most Affected by AI, the gap between junior and senior workers in Veterans vs. Juniors, and the gap between white-collar forecasts and actual data in Will AI Eliminate White-Collar Jobs?
The short answer: jobs that resist AI share three principles — (1) physical presence, (2) high-responsibility judgment that people legally and ethically carry, and (3) creativity combined with relationships built on accumulated trust. A fourth category is also growing: the people who build, secure and put AI to work. For the numbers, we use the 2025–2035 employment projections published by the U.S. Bureau of Labor Statistics (BLS) on August 27, 2026. They are one of the few public datasets that show projected growth, median pay and the education or training typically needed to enter each occupation in a single table. They are U.S. figures and do not transfer directly to other countries.
Four Categories AI Struggles to Replace
— Sharpen the picture of the growing side, not just the losses
Figures are BLS 2025→2035 U.S. employment projections (average for all occupations: +3.5%).
Globally, the WEF expects a net gain of about 78 million jobs by 2030 (about 170 million created − about 92 million displaced).
1. Look at Growing Jobs, Not Job-Loss Headlines — WEF and BLS Data
Career content is flooded with "jobs AI will replace" rankings. The core message of the WEF Future of Jobs Report 2025 points the other way. Based on a survey of more than 1,000 employers (together employing over 14 million people), it expects that between 2025 and 2030 about 170 million jobs — 14% of current employment — will be created, about 92 million (8%) displaced, for a net gain of about 78 million (7%). Work is not shrinking; its mix is changing.
So where is the growth? The WEF separates two measures. By growth rate (%), the fastest-growing roles are technology jobs such as big data specialists, fintech engineers, AI and machine learning specialists, and software and application developers. By absolute numbers, the largest growth is in frontline roles — farmworkers, delivery drivers, construction workers, salespersons and food processing workers — plus care roles such as nursing professionals, social work and counselling professionals, and personal care aides, as well as secondary and tertiary teachers. "Fastest-growing by percentage" and "adding the most people" are different lists, and it pays to keep them apart.
Narrowing to the U.S., the BLS projections point the same way. From 2025 to 2035, the fastest-growing occupational groups are healthcare support (+13.3%) and healthcare practitioners (+8.0%), which together account for almost a third of all new jobs. The fastest-declining group is office and administrative support (−4.0%, about 752,000 fewer jobs), and BLS cites automation, including AI, as a reason.
2. Three Principles of AI-Resistant Work — Physical, Responsibility, Relationships
AI-resistant jobs are not scattered at random. Three shared principles show up. Here we organize them against the reasons BLS gives for each occupation's projection and against Anthropic's March 2026 measurement of which tasks AI is actually being used for, based on Claude usage data.
Three principles that make work hard to replace
A fourth category — the people who run the AI — is growing too. See §6.
Key point: the more principles a job meets at once, the harder it is to replace (surgeon = physical + responsibility; executive = responsibility + relationships).
Meeting these principles does not mean being untouched. For radiologists, BLS writes that AI tools are making their workflow more efficient, which holds back demand in physicians' offices and hospitals (the projection is still +3.4%). Even in high-responsibility jobs, some tasks move to AI and the same work gets done by fewer people. What lasts is not the job title but the judgment and responsibility inside it.
3. Category 1: Physical Jobs — Healthcare, Care Work and the Trades
This is where BLS's fastest-growing occupations cluster. Generative AI is absorbing text and data work quickly, but running wiring in a cramped attic, lifting an older person from bed to wheelchair, or examining and treating a patient is a different problem. Aging populations also push up demand for healthcare and care. BLS additionally points to rising electricity demand (including AI-related demand such as data centers) and puts solar panel installers and wind turbine technicians among the fastest-growing occupations.
The numbers from the BLS 2025–2035 projections: nurse practitioners grow +41.0%, the fastest of all occupations (median pay $132,300). Home health and personal care aides grow +18.1%, adding about 850,000 jobs (median pay $35,800). Physical therapists +11.9%, electricians +9.2% ($63,190), HVAC technicians +10.9%, plumbers +6.8%. All exceed the +3.5% average for all occupations.
A point that surprises many: BLS lists the typical entry path for electricians and plumbers as a high school diploma plus an apprenticeship. These jobs do not assume a college degree, and their median pay is above that of general office clerks ($45,010, −6.0%). Pay varies widely by region and experience, though, and apprenticeships take several years. The shift described in Will AI Eliminate White-Collar Jobs? — office support shrinking while frontline and care work grows — shows up most clearly in these numbers.
4. Category 2: High-Responsibility Judgment — Clinicians, Lawyers, Management
These are jobs where the structure of "AI proposes → a human gives final approval" itself needs a person. In fields where legal and ethical responsibility cannot be shifted onto AI, tasks get more efficient but the position of the person who makes the final call remains.
The BLS numbers: physician assistants +21.1% ($135,880), medical and health services managers +24.2% ($123,860), lawyers +4.7% ($159,670), airline pilots +8.3% ($232,140), accountants and auditors +5.0% ($83,680). Lawyers grow at about the average, and BLS does not list AI as a factor for them. In the same legal field, however, paralegals are projected at −0.3%, with "research and document preparation increasingly automated, including by AI" as the reason. Within one industry, the side that holds judgment and responsibility and the side that does preparatory work move in different directions.
Our view: a professional license plus long practical experience will remain a strong combination. But the preparatory work newcomers have traditionally been given is exactly what moves to AI most easily, so even in licensed professions the way people gain experience in their first years is changing. Data on how the impact differs between junior and senior workers is in Veterans vs. Juniors.
5. Category 3: Creativity × Relationships — Counselors, Education, Direction
This category is about understanding people and accumulating trust over time. The more AI absorbs tasks with a right answer, the larger the share of conversations without one that people handle.
Five people-and-creativity roles
What they share: work that deals in trust that AI finds hard to replicate — relationships, organizational understanding and interpersonal sensitivity built over years.
The educator numbers show how to read this category. The WEF expects education roles to grow mainly where working-age populations are expanding, largely in lower-income economies. BLS sees U.S. elementary teachers at −0.4% and secondary teachers at −0.2%, roughly flat. Being hard for AI to replace and growing in number are two different things — the latter is decided by population, budgets and other factors unrelated to AI.
6. Category 4: The People Who Run the AI — Moving Opposite to Pure Coding Jobs
These are the jobs that build, secure and integrate AI. In the WEF's growth-rate ranking, big data specialists, fintech engineers, AI and machine learning specialists, and software and application developers are at the top. BLS projects strong growth too: data scientists +34.6% (third fastest of all occupations, $120,230), computer and information research scientists +21.8%, and information security analysts +21.0% ($129,180).
What you cannot miss is the opposite movement inside IT. In the BLS projections, software developers grow +10.2% but computer programmers shrink −7.3%. For the latter, BLS cites programming work moving to other occupations and further automation, including with AI. Anthropic's usage data also found Claude being used for 75% of programmers' tasks. As tools like Cursor and Claude Code spread, the value of writing code on its own falls, while the value of deciding what to build and using AI to design and run systems rises. That mirrors the pattern in Veterans vs. Juniors.
7. 15 Growing Jobs — BLS Projections, Pay and Entry Requirements
Across the four categories, here are 15 occupations that grow in the BLS 2025–2035 projections. Pay is the BLS 2025 median (U.S., excluding the self-employed), and entry requirements are the education, experience and training BLS lists as typically needed.
| Category | Occupation | Median pay | 2025→35 | Typical entry path (BLS) |
|---|---|---|---|---|
| Physical | Nurse practitioners | $132,300 | +41.0% | Master's degree |
| Physical | Home health and personal care aides | $35,800 | +18.1% | High school diploma + short-term on-the-job training |
| Physical | Physical therapists | $102,760 | +11.9% | Doctoral or professional degree |
| Physical | HVAC mechanics and installers | $61,010 | +10.9% | Postsecondary nondegree award + long-term on-the-job training |
| Physical | Electricians | $63,190 | +9.2% | High school diploma + apprenticeship |
| Physical | Plumbers, pipefitters and steamfitters | $63,800 | +6.8% | High school diploma + apprenticeship |
| Judgment | Medical and health services managers | $123,860 | +24.2% | Bachelor's degree + under 5 years of related experience |
| Judgment | Physician assistants | $135,880 | +21.1% | Master's degree |
| Judgment | Airline pilots | $232,140 | +8.3% | Bachelor's degree + under 5 years of related experience + moderate-term training |
| Judgment | Lawyers | $159,670 | +4.7% | Doctoral or professional degree |
| Creative × people | Substance abuse and mental health counselors | $59,350 | +18.4% | Master's degree + internship/residency |
| Creative × people | Clinical and counseling psychologists | $100,580 | +11.7% | Doctoral degree + internship/residency |
| AI operators | Data scientists | $120,230 | +34.6% | Bachelor's degree |
| AI operators | Computer and information research scientists | $140,300 | +21.8% | Master's degree |
| AI operators | Information security analysts | $129,180 | +21.0% | Bachelor's degree + under 5 years of related experience |
Source: U.S. Bureau of Labor Statistics (BLS), Employment Projections 2025–35, Table 1.2 (published August 27, 2026). Average for all occupations: +3.5%; median pay: $50,980.
The common pattern: most require a license or degree, or several years of apprenticeship or experience. There are hardly any occupations you can enter quickly with no experience that are also growing fast. Home health aides come closest — the entry bar is low, but so is median pay at $35,800. To get into a growing occupation, look at the entry requirements together with how long and how much it takes to meet them. The survival strategy is to move from where you are now to one of these categories by the shortest route (next section).
8. Four Moves From Your Current Career to the Growing Side
To keep this concrete, assume you currently work in an office, sales or entry-level programming role. Here are four specific moves toward the growing side.
Four moves to the growing side
Common thread: deliberately move from tasks AI shrinks to tasks it struggles to shrink.
You can move at any age, but for occupations that require a degree or apprenticeship, moving early leaves more options.
Advice for new graduates: in the BLS projections, general office clerks (−6.0%), customer service representatives (−5.3%) and computer programmers (−7.3%) are on the declining side. Taking such a job is not a mistake in itself, but have a plan from day one to connect what you learn to judgment, people skills or AI use. Conversely, nursing and medical technology, electrical work, caregiving, and data and security roles grow faster than average in the BLS projections. Choosing a career on the assumption that office work ranks above hands-on work can make you overlook the growing side.
Summary
Job-loss stories dominate, but the WEF expects a net gain of about 78 million jobs by 2030 (about 170 million created − about 92 million displaced). Work is not disappearing; its mix is changing. AI-resistant jobs share three principles: (1) physical presence (healthcare, care work, the trades), (2) high-responsibility judgment (clinicians, lawyers, management), and (3) creativity × relationships (counselors, education, executives). A fourth category, the people who run the AI (data scientists, information security, researchers), is growing as well.
In the BLS 2025–2035 projections, nurse practitioners (+41.0%), data scientists (+34.6%), information security analysts (+21.0%), home health aides (+18.1%) and electricians (+9.2%) beat the +3.5% average, while general office clerks (−6.0%) and computer programmers (−7.3%) decline. Most growing occupations are entered through a license, degree or apprenticeship. The four moves to the growing side are (1) become the AI user, (2) go deep in an industry, (3) reconsider physical work, and (4) invest in relationships. Check the entry requirements and timeline, then move early.
Related reading: Will AI Eliminate White-Collar Jobs?, The Future of Sales Jobs, Veterans vs. Juniors, 15 Jobs Most Affected by AI and What Is a Forward Deployed Engineer?
FAQ
Q. I work in an office role (accounting, HR, sales admin). What should I do now?
A. In the BLS projections, general office clerks (−6.0%) and secretaries and administrative assistants (−6.0%) are on the declining side. They will not vanish overnight, but the earlier you move, the more options you have. The shortest route is moving into an AI adoption or process-improvement role inside your current company. Experience with spreadsheets, accounting software and internal systems carries over directly as business knowledge. If you leave, work backward from the entry requirements of a growing field you can enter with a bachelor's degree and a few years of experience, such as information security (+21.0%).
Q. Is it too late to change careers after 40?
A. The relationship category rewards life and work experience. Care roles (caregiving, counseling) also value life experience, but BLS lists a master's degree plus internship as the entry path for counselors, so plan for time to retrain. Electrician and plumber apprenticeships take several years. Moving toward the AI-user side can start inside your current job, so that is where we suggest people in their 40s begin.
Q. How should I advise my children on careers?
A. In the BLS projections, fields growing well above average include healthcare (nurse practitioners, physician assistants, physical therapists), data and information security, and mental health. Lawyers and accountants grow at about the average, and BLS does not cite AI as a factor for them. Whatever the path, encourage "field expertise × the ability to use AI" from the start. Conversely, jobs centered on routine clerical processing and jobs that are only about writing code are on the declining side of the BLS projections.
Q. Is switching from office work to a skilled trade realistic?
A. The BLS numbers make it a real option. Median pay for electricians ($63,190, +9.2%) and plumbers ($63,800, +6.8%) is above general office clerks ($45,010, −6.0%), and entry is a high school diploma plus apprenticeship. But apprenticeships take years, the work is physically demanding, and pay varies widely by region and experience. It is neither a demotion nor an easy way out — it is one path to choose after checking the conditions.
Q. Is being great at using AI enough on its own?
A. Not on its own. The more widespread AI skills become, the less they set you apart. What lasts is a combination: AI skills × deep industry knowledge, AI skills × licensed judgment, or AI skills × relationships. Combining several axes is harder to replace than relying on AI alone, industry alone or relationships alone.
