Research · Retirement Hub · June 2026
Which Jobs Still Exist at Age 65 in 2040? — A BLS + AI-Exposure Analysis
If you're 50 today, the BLS Employment Projections through 2032 plus the Eloundou et al. AI exposure scores together imply a fairly small list of occupations likely to be both growing and resistant to LLM substitution when you turn 65. This piece walks the intersection — and the much shorter list of late-career-friendly jobs inside it.
By Retirement Hub — AI Impact on Retirement · Updated 2026-06-21 · Educational only — not financial, tax, or investment advice.
The Bureau of Labor Statistics releases ten-year Employment Projections every two years. The current edition (released September 2023, covering 2022–2032) projects total US employment to grow about 4.7 million jobs, or 2.8% over the decade — well below the 7.6% growth of the prior decade. Behind that headline is enormous variation: healthcare, computer occupations, and renewable energy installation are projected to grow 20–40%; office and administrative support, retail sales, and many production occupations are projected to shrink.
Overlay the 2023 Eloundou, Manning, Mishkin, Rock paper 'GPTs are GPTs' — which scored every detailed occupation on the share of its task content that an LLM can meaningfully accelerate — and the BLS growth projections look different. Many of the occupations BLS projects to grow (software developers, market research analysts, financial analysts) score high on Eloundou exposure. Many of the occupations Eloundou flags as low-exposure (construction, healthcare practitioners, in-person services) are physically demanding in ways that disadvantage 60-something workers.
The intersection of three filters — projected employment growth through 2032+, low LLM exposure, and physical demands manageable by a typical 60+ worker — is what this piece focuses on. The list is short. It is not a forecast for any individual; it is a planning tool for workers considering whether to retrain into a new occupation in their 50s. Pair this with the AI displacement → retirement age calculator to estimate what early-exit scenarios mean for your number.
Occupations growing through 2032 AND low LLM exposure — the intersection
| Occupation | BLS 2022–2032 growth | Median wage (BLS 2023) | AI exposure (Eloundou E1) | Late-career fit (60+ worker) |
|---|---|---|---|---|
| Registered Nurses | +6% (177k jobs) | $86,070 | 0.10 (low) | Moderate — physical demands |
| Nurse Practitioners | +45% (118k) | $126,260 | 0.15 (low-mid) | Good — less physically demanding |
| Physical Therapists | +15% (37k) | $99,710 | 0.12 (low) | Moderate — physical |
| Mental Health Counselors | +18% (43k) | $53,710 | 0.18 (low-mid) | Excellent — sit-down clinical |
| Clergy | +3% (8k) | $58,920 | 0.08 (very low) | Excellent |
| Carpenters | +2% (24k) | $56,350 | 0.06 (very low) | Poor — physical |
| Electricians | +6% (49k) | $61,590 | 0.07 (very low) | Poor — physical |
| Plumbers | +2% (10k) | $61,550 | 0.07 (very low) | Poor — physical |
| HVAC Mechanics | +6% (24k) | $57,300 | 0.08 (very low) | Moderate |
| Home Health Aides | +22% (765k) | $33,530 | 0.10 (low) | Poor — physical + low wage |
| Personal Care Aides | +25% (820k) | $33,530 | 0.10 (low) | Poor — physical + low wage |
| Dental Hygienists | +7% (15k) | $87,530 | 0.20 (low-mid) | Moderate |
| Chefs and Head Cooks | +5% (8k) | $58,920 | 0.05 (very low) | Poor — physical + hours |
| Compliance Officers (regulatory inspection) | +5% (16k) | $75,670 | 0.32 (mid) | Excellent |
| Funeral Directors | +5% (1.5k) | $60,330 | 0.05 (very low) | Excellent |
| Massage Therapists | +18% (33k) | $55,310 | 0.04 (very low) | Poor — physical |
| Speech-Language Pathologists | +19% (33k) | $89,290 | 0.15 (low-mid) | Excellent — sit-down clinical |
| Occupational Therapists | +12% (17k) | $96,370 | 0.13 (low-mid) | Moderate |
| Substance Abuse Counselors | +18% (78k) | $53,710 | 0.18 (low-mid) | Excellent |
Sources: BLS Employment Projections 2022–2032 (Sept 2023 release); Eloundou et al. (2023) Table 3 occupation-level E1 scores; BLS Occupational Employment and Wage Statistics 2023. 'Late-career fit' is Retirement Hub's qualitative judgment based on physical demands, scheduling flexibility, and the BLS occupation profile. Wages are national medians; localized wages vary substantially.
Why the list is dominated by healthcare
Two of the top three projected-growth occupational groups in the BLS data are healthcare practitioners (RNs, NPs, PAs, therapists) and healthcare support (home aides, personal care). Both score low on Eloundou LLM exposure because their tasks are physical, interpersonal, and licensed-liability. The high-wage healthcare practitioner roles also tend to score well on late-career fit because the physical demands are manageable.
The downside: most healthcare practitioner roles require multi-year credentialing. An NP credential typically takes 6–8 years from a non-nursing start; a speech-language pathology master's is 2–3 years post-bachelor's. For a 50-something considering a mid-career switch, the runway question is real — and the BLS Occupational Outlook Handbook entry for each role lists the typical credentialing path.
Skilled trades grow, but late-career fit is poor
Electricians, plumbers, and HVAC mechanics all show BLS-projected growth and very low LLM exposure (mechanical work is not what current LLMs do well). But every trade ranks 'poor' or 'moderate' on late-career fit because the physical demands — ladder work, weight-bearing, crawling in confined spaces — get progressively harder past 55.
The exception is supervisory and inspection work. An electrician with 25 years of field experience often transitions to inspection, code consulting, or general contracting in their late 50s and stays employable through their 60s. The trade gives you a credential pathway; the credential lets you exit the physical work without leaving the field.
The 'compliance' anomaly
Compliance officers and regulatory inspectors are an interesting anomaly: BLS projects modest growth, Eloundou scores them mid-exposure (0.32), but the late-career fit is excellent. The reason is that AI accelerates the *routine* parts of compliance (document review, transaction screening) while leaving the *judgment* parts (whistleblower investigations, regulatory negotiation, board-level reporting) intact. Workers in the field who move up the seniority curve into judgment-heavy work see relatively little near-term displacement and high late-career employability.
The same pattern applies to insurance adjusters, financial examiners, and many regulatory specialists. The mid-exposure scores in Eloundou tend to overstate the displacement risk for senior workers in these fields.
Get the retirement playbook in your inbox.
One weekly email with new calculators, source updates, and AI-impact research. Free, no spam, unsubscribe anytime.
Frequently asked questions
If I'm a software developer, am I in trouble?+
Mixed. Software developers score 0.40+ on Eloundou exposure (mid-high), but BLS projects 25% growth through 2032. The signal from firm-level studies (GitHub 2023, Microsoft 2024) is that productive senior developers using AI tools are getting *more* productive, not less employable. Junior coding work appears to be where displacement is concentrated. For a 50-something developer, the planning question is whether your current employer values the architectural and judgment work LLMs can't yet do — not whether the field overall is shrinking.
Why is the list so short?+
Because the three filters compound. Most occupations fail at least one — either they're shrinking (retail clerks, office support), they're high LLM exposure (analysts, writers, paralegals), or they're too physically demanding for typical 60+ workers (construction, warehousing). The intersection is small by construction, and that's the point: late-career-resilient occupations are scarce, which is why the AI-displacement scenario in the retirement-age calculator matters so much for planning.
What about jobs that don't exist yet?+
By 2040, some currently-tiny occupations will be large. AI workflow integration specialists, prompt-engineering consultants, and AI-output auditors are real and growing. But the BLS does not yet track them as separate occupations, so the data is too thin for inclusion. Historically the BLS adds new SOC codes 10–15 years after a job category becomes large enough to measure; expect new AI-adjacent SOC codes in the 2030s.
Should I plan to work to 70 if my job is on this list?+
Yes, conditional on health. The list is filtered for late-career fit precisely because the value of working an extra five years is enormous: each year of additional Social Security deferral past FRA adds 8% to your monthly benefit, plus you avoid five years of drawdown. See the SS claim-age calculator and the retirement gap analysis.
Sources
- BLS Employment Projections 2022–2032 (September 2023 release)
- BLS Occupational Outlook Handbook
- BLS Occupational Employment and Wage Statistics (OEWS) 2023
- Eloundou, Manning, Mishkin, Rock — GPTs are GPTs (March 2023)
- BLS Employee Tenure Survey 2024
- Goldman Sachs — Potentially Large Effects of AI on Economic Growth (March 2023)
Related on Retirement Hub
Sibling project
Building a retirement plan with ChatGPT or Claude?
AIPromptsHub.cohas 40+ free prompt templates for financial planning research, retirement Q&A, and AI-assisted budgeting — built by the same team.