You've seen the headlines. Goldman Sachs, McKinsey, OpenAI — everyone has a number. But what does the actual labor data say in 2026, by college major? Let's look at what's real, what's hype, and what to do about it.
The headlines, in context
Three numbers shaped most people's anxiety. Let's translate each one accurately.
Goldman Sachs: "300 million jobs exposed to AI"
This 2023 number is real but widely misread. The Goldman report said 300 million full-time-equivalent jobs globally have some exposure to AI automation — meaning AI could perform some of the tasks. Goldman's actual estimate of jobs likely to be fully replaced was around 7% of the U.S. workforce — far smaller, and concentrated in specific occupations.
OpenAI / University of Pennsylvania: "80% of jobs affected"
The Eloundou et al. paper from OpenAI and Penn (2023) said 80% of U.S. workers will see at least 10% of their tasks affected by GPT-class models. That's a measure of task exposure, not job loss. The very same paper estimated only 19% of workers would see more than half their tasks affected.
McKinsey: "Up to 30% of work hours automated by 2030"
McKinsey's 2023 model projected automation of 27-30% of all U.S. work hours by 2030 — but the same report concluded this would create more jobs than it destroyed in healthcare, STEM, and skilled trades, while reducing demand in office support and customer service.
The honest synthesis: AI is real, the disruption is real, and the impact is highly uneven by occupation and major. Some categories are getting cut. Others are growing faster because of AI.
Match Your Strengths to an AI-Resilient Major →
Don't pick a major based on headlines. Pick one based on the actual 2026 labor data — matched to how you think.
AI replacement risk by college major
Here's what the data shows, organized by major, using a synthesis of Brookings (AI Exposure Index), OECD (Skill Use and AI), and BLS occupational projections.
| Major | AI exposure | Job growth 2024-2034 (BLS) | Real-world signal |
|---|---|---|---|
| Nursing (RN/BSN) | Low | +6% (faster than average) | Severe shortage; AI assists, doesn't replace |
| Civil engineering | Low-medium | +6% | Infrastructure boom; AI accelerates design only |
| Computer science | Medium | +17% | Junior coding compressed; senior roles up |
| Accounting | Medium-high | +6% | CPA holders rising; bookkeepers declining |
| Communications / generic marketing | High | +6% (but quality varies) | Junior content roles compressed |
| Paralegal / legal studies | High | +1% (BLS revised down 2025) | AI doing first-pass document review |
| Education (K-12) | Low | +1-3% | AI tools used by teachers, not replacing them |
| Skilled trades | Very low | +5-11% (varies by trade) | Severe shortage; physical work unautomatable |
| Mechanical/Electrical engineering | Low-medium | +10% | Manufacturing reshoring drives demand |
| Psychology (BA only) | Medium | Limited entry-level | Therapy roles strong; research roles compressed |
| Business administration (general) | Medium-high | +5% | Generic credential, weakening signal |
| Graphic design | High | -4% (BLS) | Generative AI directly replaces entry work |
The pattern: licensed work, hands-on work, and senior judgment work are growing. Generic credentials and entry-level information processing are compressing.
What's actually happening in 2026 hiring
Three patterns are visible in the labor data through Q1 2026:
1. The "experience cliff" is widening
Junior knowledge-work roles are getting harder to land. According to Stanford's 2025 AI Index, entry-level postings in software, marketing, and legal research declined 14-21% in 2024-2025, while mid-senior postings in those same fields grew 7-11%. Translation: AI is eating the rungs of the career ladder, not the top.
2. Healthcare and trades are hiring fast
BLS Job Openings and Labor Turnover data shows healthcare openings ran above 1.8 million through 2025, and skilled trades have a structural shortage now estimated at 500,000+ workers. These are the careers where the labor market remains tight.
3. The wage signal is loud
Where AI raises productivity in scarce-credential fields, wages rise. Where AI raises productivity in abundant-credential fields, headcount falls. (See the BLS data summarized in our AI productivity article.)
What to do about it
If you're choosing a major
Use a 4-question filter:
- Is this work licensed or scarcity-protected? RN, PE, CPA, JD, MD, electrician — all create durable wage premiums.
- Does the work require physical presence or human trust? Therapy, teaching, healthcare, trades, social work.
- Is the credential the floor of the job, not the whole job? Engineering and nursing get you in the door; the work itself takes years to master and AI can't shortcut that.
- Is BLS projecting growth, not decline? Cross-check at bls.gov/ooh.
If you've already chosen
Specialize aggressively, get the licensing if it exists in your field, and learn the AI tools used in your domain. The 2024 BCG study (Dell'Acqua et al.) showed clearly: AI-fluent specialists outperform AI-fluent novices by huge margins. Domain expertise still wins.
See Which Majors Match Your Strengths →
Our science-backed quiz uses 60 research-driven questions plus current BLS data to surface majors that fit you and weather the AI shift.
Frequently Asked Questions
Will AI replace 80% of jobs?
No. The "80%" figure from the Eloundou OpenAI/Penn paper measured task exposure, not job loss. The same paper estimated only 19% of workers will see more than half their tasks meaningfully affected. Most jobs change. Few disappear.
Which jobs are AI definitely cutting in 2026?
Generic content writing, junior paralegal work, entry-level graphic design, basic bookkeeping, first-tier customer support, and routine data entry. These are the categories where headcount is observably declining at large employers.
Is computer science still safe?
Senior CS is safer than ever. Junior CS is harder than it was. Outcomes increasingly depend on skill depth — systems design, distributed systems, AI/ML expertise, security — and not just being able to write code that AI can also write.
What about healthcare and trades?
Both are AI-resistant for different reasons. Healthcare has trust and licensing moats; trades have physical-world moats. Both are hiring aggressively in 2026 with rising wages.
Is a college degree still worth it in the AI age?
For most paths, yes — but the "worth it" calculation depends entirely on which major and which credential. Read our deeper analysis on this question.
Sources
- Goldman Sachs Economic Research. "The Potentially Large Effects of Artificial Intelligence on Economic Growth." 2023.
- Eloundou, Manning, Mishkin, Rock. "GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models." OpenAI/Penn, 2023.
- McKinsey Global Institute. "Generative AI and the future of work in America." 2023, 2025 update.
- Brookings Institution. "AI Exposure Index" (Muro, Whiton, Maxim).
- OECD. "AI and the Future of Skills" series, 2024-2025.
- Bureau of Labor Statistics. Employment Projections 2024-2034. bls.gov/emp.
- Bureau of Labor Statistics. Job Openings and Labor Turnover Survey, 2025.
- Stanford HAI. "AI Index Report 2025."
- Dell'Acqua et al. "Navigating the Jagged Technological Frontier." Harvard Business School, 2023.