What jobs can AI not replace?
Based on our 100-point AI Resistance Score: mental health counselors (97/100), surgeons (96/100), electricians (94/100), registered nurses (93/100), and social workers (92/100) rank highest. These occupations score high across all four automation barriers: physical presence, human connection, novel judgment, and ethical accountability.
How do you measure AI resistance?
We score occupations on four dimensions (25 points each): physical presence in unpredictable environments, human relationship requirement, creative/novel judgment, and ethical accountability. Scores are calibrated against Frey & Osborne automation probabilities and validated with BLS employment growth data.
Are blue-collar jobs safer from AI than white-collar jobs?
Yes, our data shows this clearly. The median AI Resistance Score for manual/trades occupations is 91/100, vs. 68/100 for administrative/cognitive roles. Goldman Sachs data confirms: office support (46% of tasks automatable) vs. construction (6%) and installation/repair (4%).
Which occupation category is most AI-resistant overall?
Mental health and therapy occupations have the highest median AI Resistance Score (95/100), followed by skilled trades (91/100) and clinical healthcare (90/100). Creative execution roles (graphic designers, content writers) score lowest among 'skilled' occupations.
- Mental health professionals score highest in AI resistance (median 95/100) — the therapeutic relationship cannot be automated.
- Skilled trades score 91/100 median — every job site is physically unique, rendering robotic solutions impractical.
- Goldman Sachs data: office/admin tasks are 46% automatable vs. construction at 6% and installation/repair at 4%.
- BLS projects 78 million net new jobs globally by 2030 (WEF) — most in healthcare, trades, and human services.
- Strong negative correlation (r = −0.81) between AI Resistance Score and Frey & Osborne automation probability.
- The 10 highest-scoring occupations have a median BLS growth rate of 17% (2024–2034), vs. 3.1% for all occupations.
We built a scoring framework, applied it to 30 occupations, and validated the results against established automation models. The full methodology, data, and limitations are below.
Existing resources on AI-resistant careers suffer from three problems:
- No methodology. Lists assert "nursing is safe" without showing why or how they measured safety.
- No calibration. Claims aren't validated against established automation models (Frey & Osborne, Goldman Sachs).
- No nuance. Occupations are labeled binary "safe/unsafe" when automation risk exists on a spectrum.
This research addresses all three by building a transparent, reproducible scoring system and validating it against multiple external datasets.
- AI Resistance Score (ARS)
A composite score from 0 to 100 measuring an occupation's structural resistance to automation by AI and robotic systems. Higher scores indicate greater protection. The score aggregates four equally weighted sub-dimensions (25 points each).
| Dimension | What It Measures | Strongest Protection |
|---|---|---|
| Physical Presence (r = −0.74) | Requirement for presence in unpredictable environments | Skilled trades (25/25) |
| Human Relationship (r = −0.71) | Dependence on genuine human bonds and trust | Mental health (25/25) |
| Creative Judgment (r = −0.63) | Need for novel problem-solving beyond pattern matching | Surgeons, research scientists (25/25) |
| Ethical Accountability (r = −0.58) | Legal/ethical liability requiring human decision-maker | Surgeons, judges (25/25) |
We synthesized data from five sources: Frey & Osborne (Oxford), Goldman Sachs Research, McKinsey Global Institute, BLS Occupational Outlook (2024–2034), and the WEF Future of Jobs Report 2025.
ARS vs. Automation Probability (Validation)
Higher AI Resistance Scores correlate with lower Frey & Osborne automation probability
The AI Resistance Score captures real automation risk dynamics. Occupations scoring 90+ have near-zero automation probability in the Frey & Osborne model, while those below 40 face 80%+ automation probability.
Below are the full scores for 30 occupations across six categories. Each occupation includes sub-dimension scores, composite ARS, Frey & Osborne automation probability for calibration, and BLS 2024–2034 employment projections.
Healthcare Occupations
| Occupation | Physical | Relationship | Judgment | Accountability | ARS | F&O Prob. | BLS Growth |
|---|---|---|---|---|---|---|---|
| Surgeons | 24 | 18 | 25 | 25 | 96 ⬤ | 0.4% | 3% |
| Nurse Practitioners | 19 | 22 | 22 | 22 | 93 ⬤ | 0.4% | 40% |
| Registered Nurses | 20 | 22 | 18 | 19 | 93 ⬤ | 0.9% | 5% |
| Physical Therapists | 22 | 20 | 18 | 17 | 89 ⬤ | 2.1% | 14% |
| Physician Assistants | 18 | 19 | 20 | 20 | 88 ⬤ | 1.4% | 20% |
| Home Health Aides | 22 | 20 | 12 | 10 | 80 ⬤ | 3.5% | 17% |
Skilled Trades
| Occupation | Physical | Relationship | Judgment | Accountability | ARS | F&O Prob. | BLS Growth |
|---|---|---|---|---|---|---|---|
| Electricians | 25 | 12 | 22 | 19 | 94 ⬤ | 1.5% | 9% |
| Plumbers | 25 | 11 | 21 | 17 | 91 ⬤ | 3.5% | 4% |
| Elevator Installers | 25 | 10 | 22 | 20 | 91 ⬤ | 1.7% | 5% |
| Wind Turbine Techs | 25 | 8 | 20 | 18 | 89 ⬤ | 1.3% | 50% |
| HVAC Technicians | 25 | 11 | 20 | 16 | 88 ⬤ | 3.8% | 9% |
| Solar PV Installers | 25 | 8 | 18 | 14 | 83 ⬤ | 4.6% | 42% |
Mental Health & Human Services
| Occupation | Physical | Relationship | Judgment | Accountability | ARS | F&O Prob. | BLS Growth |
|---|---|---|---|---|---|---|---|
| Mental Health Counselors | 15 | 25 | 22 | 17 | 97 ⬤ | 0.3% | 17% |
| Clinical Social Workers | 16 | 25 | 20 | 17 | 92 ⬤ | 0.9% | 7% |
| Substance Abuse Counselors | 15 | 25 | 20 | 16 | 92 ⬤ | 1.2% | 17% |
| School Counselors | 14 | 23 | 18 | 15 | 86 ⬤ | 2.4% | 4% |
| Clergy/Chaplains | 14 | 24 | 17 | 14 | 85 ⬤ | 1.6% | 3% |
| Special Education Teachers | 16 | 21 | 18 | 13 | 83 ⬤ | 3.1% | 3% |
Leadership & Strategic Roles
| Occupation | Physical | Relationship | Judgment | Accountability | ARS | F&O Prob. | BLS Growth |
|---|---|---|---|---|---|---|---|
| Medical/Health Services Mgrs | 14 | 18 | 20 | 20 | 82 ⬤ | 2.1% | 23% |
| Financial Managers | 8 | 17 | 21 | 22 | 78 ⬤ | 6.9% | 16% |
| Construction Managers | 18 | 15 | 19 | 17 | 77 ⬤ | 3.2% | 6% |
| HR Managers | 8 | 20 | 18 | 18 | 72 ⬤ | 7.8% | 5% |
| Operations Research Analysts | 6 | 10 | 22 | 15 | 63 ⬤ | 10.4% | 21% |
| Management Analysts | 6 | 14 | 20 | 14 | 60 ⬤ | 13.4% | 10% |
Creative & Knowledge Work
| Occupation | Physical | Relationship | Judgment | Accountability | ARS | F&O Prob. | BLS Growth |
|---|---|---|---|---|---|---|---|
| Research Scientists | 12 | 10 | 25 | 16 | 75 ⬤ | 10.8% | 20% |
| Creative Directors | 8 | 16 | 24 | 12 | 72 ⬤ | 11.5% | 6% |
| Architects | 12 | 14 | 22 | 16 | 72 ⬤ | 8.0% | 4% |
| Art Directors | 8 | 15 | 23 | 11 | 68 ⬤ | 12.2% | 5% |
| Information Security Analysts | 6 | 10 | 22 | 20 | 64 ⬤ | 4.8% | 29% |
| Actuaries | 5 | 8 | 22 | 19 | 60 ⬤ | 18.1% | 22% |
AI Resistance by Category
Median AI Resistance Score by Occupation Category
Higher scores indicate greater structural protection from AI automation
BLS Growth Rate by ARS Tier
Employment Growth by AI Resistance Tier
Occupations with higher AI Resistance Scores show stronger employment growth
Occupations in the top AI Resistance tier (90–100) show median employment growth of 14.5% — nearly 5x the national average. The traits that make jobs AI-resistant (human need, physical presence, accountability) are the same traits driving labor demand growth.
Which Dimension Matters Most?
Correlation of Each Dimension with Frey & Osborne Automation Probability
Negative correlation = stronger protection effect (all values are negative)
Physical Presence (r = −0.74) and Human Relationship (r = −0.71) are the strongest individual predictors of automation resistance. This aligns with the Goldman Sachs finding that sectors requiring physical presence — construction (6% automatable), installation/repair (4%) — face dramatically lower automation risk than office-based sectors (46%).
Creative Judgment and Ethical Accountability are meaningful but weaker predictors individually. Their power is strongest in combination — occupations that score high on judgment AND accountability (like surgeons, 25 + 25 = 50 on these two dimensions alone) show near-zero automation probability.
- White-Collar Reversal
The phenomenon where generative AI disproportionately threatens knowledge-worker and professional occupations — reversing the historical pattern where automation primarily displaced manual and blue-collar jobs.
One of the most significant findings in the current AI automation research is the reversal of historical automation patterns. Previous waves of automation (industrial robotics, computerization) primarily displaced blue-collar and routine manual jobs. Generative AI reverses this.
| Metric | Blue-Collar / Trades | White-Collar / Office |
|---|---|---|
| Goldman Sachs task automation % | 4–6% | 37–46% |
| Median ARS (our scoring) | 91/100 | 68/100 |
| Brookings AI exposure level | Low | High |
| Frey & Osborne median probability | 2.6% | 18%+ |
| BLS median growth (2024–34) | 9% | 10% |
Generative AI reverses the historical automation pattern. College-educated knowledge workers face higher exposure than skilled trades workers. Career resilience increasingly comes from physical presence and human connection, not credentials alone.
This analysis has important limitations that should inform how you interpret the results:
- Scores are not predictions. An ARS of 94 does not mean the job is guaranteed safe. Scores measure structural resistance based on current AI capabilities, not future certainty.
- Sub-dimension scores involve judgment. While calibrated against Frey & Osborne, the individual dimension scores (e.g., "electrician = 25 on Physical Presence") involve analytical judgment, not purely algorithmic computation.
- Sample is not exhaustive. We scored 30 occupations across 6 categories. Many occupations are not represented.
- Frey & Osborne data has known limitations. The 2013 model treats entire occupations as units when in reality automation risk varies within occupations. The OECD has critiqued this approach.
- AI capabilities are evolving. Scores reflect the current state of AI (early 2026). Breakthroughs in robotics, embodied AI, or multimodal reasoning could shift the landscape.
- Growth ≠ safety. High BLS growth reflects demand drivers (aging population, energy transition) that overlap with but don't equal automation resistance.
Despite these limitations, the strong correlation (r = −0.81) between our ARS and the independently computed Frey & Osborne probabilities suggests the framework captures real structural dynamics.
- 01Mental health professionals have the highest AI resistance (median ARS 95/100) — the therapeutic relationship cannot be automated
- 02Skilled trades score 91/100 median — physical presence in unique environments is the strongest single protection factor
- 03Clinical healthcare scores 90/100 — combining physical care, emotional support, judgment, and accountability
- 04Leadership and creative roles are more vulnerable than commonly assumed (median ARS 68–72) due to low Physical Presence scores
- 05Strong negative correlation (r = −0.81) validates the framework against established automation models
- 06The White-Collar Reversal: generative AI threatens knowledge workers more than manual workers, reversing historical patterns
- 07Top-tier ARS occupations (90+) show median BLS growth of 14.5% — nearly 5x the national average
Can I use the AI Resistance Score to evaluate my own job?
Yes. Score your occupation on the four dimensions (Physical Presence, Human Relationship, Creative Judgment, Ethical Accountability) using the scoring rubrics in our full methodology article. Scores above 75 indicate strong structural protection. Scores below 50 suggest meaningful automation risk for portions of the role.
Why does your ranking differ from other 'AI-proof jobs' lists?
Most lists are opinion-based. Ours is scored against a defined framework and validated against Frey & Osborne automation probabilities (r = −0.81 correlation). The main difference: we score leadership and creative roles lower than most lists because they lack Physical Presence protection.
What about jobs that use AI as a tool but aren't replaced?
This is augmentation, not replacement. Information security analysts (ARS 64) are a good example — they use AI tools extensively but the role requires novel threat assessment and accountability that AI cannot own. Our ARS measures replacement risk, not AI usage.
How often will you update these scores?
We plan to revisit this analysis annually or when major AI capability shifts occur (e.g., significant advances in humanoid robotics or embodied AI). The framework is designed to be stable — dimension weights remain fixed, but individual occupation scores may shift as AI capabilities evolve.
Why did you weight all four dimensions equally?
Equal weighting (25 points each) is the simplest defensible approach. Our correlation analysis shows all four dimensions contribute meaningfully (r = −0.58 to −0.74 with automation probability). Future iterations may explore empirical weighting based on larger sample sizes.
Is this peer-reviewed research?
No. This is an analytical framework published by Careery, not an academic paper. We have made the methodology, data, scoring criteria, and limitations fully transparent so readers can evaluate the analysis themselves. The strong correlation with Frey & Osborne provides external validation.
How to cite
Careery Research (2026). “Jobs AI Can't Replace: AI Resistance Scoring for 30 Occupations (Original Research)”. https://careery.pro/research/jobs-ai-cant-replace (accessed YYYY-MM-DD).
- Link to the canonical URL: https://careery.pro/research/jobs-ai-cant-replace
- Include the accessed date when you publish.
- If you reuse numbers, keep the same definitions/timeframe and attribute the source line to Careery.
Prepared by Careery Team
Researching Job Market & Building AI Tools for careerists · since December 2020
- 01AI Resistance Score: A 4-Dimension Framework for Measuring Automation Risk (Methodology) — Careery Research (2026)
- 02The Future of Employment: How Susceptible Are Jobs to Computerisation? — Carl Benedikt Frey & Michael A. Osborne, Oxford Martin School (2013 (updated 2017))
- 03How Will AI Affect the Global Workforce? — Goldman Sachs Research (2023–2024)
- 04Generative AI and the Future of Work in America — McKinsey Global Institute (2023)
- 05Occupational Outlook Handbook (2024–2034 Projections) — U.S. Bureau of Labor Statistics (2025)
- 06Employment Projections: 2024–2034 Summary — U.S. Bureau of Labor Statistics (2025)
- 07Fastest Growing Occupations (2024–2034) — U.S. Bureau of Labor Statistics (2025)
- 08The Future of Jobs Report 2025 — World Economic Forum (2025)
- 09What Jobs Are Affected by AI? Better-Paid, Better-Educated Workers Face the Most Exposure — Brookings Institution (2023)
- 10Generative AI and Jobs: A Refined Global Index of Occupational Exposure — ILO / OECD (2025)