AI doesn't mean every major is doomed. It means the rules for picking one have changed. Here's a 4-question framework — built from the latest McKinsey, Stanford HAI, and BLS data — to choose well.
What's actually changed about choosing a major
Three big shifts redefined the major-choice equation between 2022 and 2026:
1. The rungs of the career ladder are getting eaten
Stanford's 2025 AI Index documented a 14-21% decline in entry-level postings across software, marketing, and legal research from 2024 through Q1 2025. The work AI compresses fastest is the work most often given to recent graduates. The strongest majors are the ones that lead to roles where the work itself takes years to master, not weeks.
2. Credentials with licensing or scarcity hold value
The 2024 BCG/Harvard study (Dell'Acqua) showed that AI-fluent specialists outperform AI-fluent novices by huge margins. Translation: AI tools amplify expertise, not replace it. Licensing — RN, PE, CPA, JD, MD, electrician's license — creates a moat AI can't cross because the law won't let it.
3. Generic credentials are weakening
BLS wage data through May 2024 (released 2025) shows median wages for software developers, registered nurses, and electricians rose 5.9-7.1% year-over-year. Median wages for general office clerks and entry-level paralegals rose only 2.1-2.4%. The gap is widening.
The 4-question framework
Run any major you're considering through these four questions. The more "yes" answers, the stronger the major in 2026.
Question 1: Does this major lead to work that's licensed or scarcity-protected?
Licensing is the cleanest moat against AI. Nursing, engineering (toward PE), accounting (toward CPA), law, medicine, the trades — all carry mandatory credentials. The state literally won't let an AI do the job.
Scarcity-protection works similarly. Software at senior levels, specialty medicine, university teaching — fields where the supply of people who can do the work is constrained by training years.
Strong yes: Nursing, all engineering disciplines, accounting (CPA path), pre-med, education (especially special ed and STEM), social work, the skilled trades.
Question 2: Does the work require physical presence, human trust, or judgment under uncertainty?
AI is bad at the physical world (despite robotics progress, deployment is slow and expensive in unstructured environments). AI is bad at trust — patients, clients, and students still want a human. AI is bad at judgment when stakes are high and data is incomplete.
Strong yes: Healthcare (all licensed roles), trades, teaching, therapy, social work, complex law, senior consulting, judgment-heavy management roles.
Question 3: Does the credential carry depth that AI tools amplify?
This is the AI-augmentation test. A computer science degree plus AI fluency is a compounding combination. A general communications degree plus AI fluency is just AI fluency — the credential adds nothing the AI tool doesn't already do.
Strong yes: CS, data science, engineering, finance/economics, accounting, nursing/pre-med, statistics.
Weaker: Generic business administration, generic communications, generic marketing without a specialty.
Question 4: Does the BLS project growth, and does the entry-level path still exist?
This is the practical test. Cross-check at bls.gov/ooh for the occupations your major leads to. Growth above the all-occupation 4% average is a green flag. Decline is a red flag. Pay attention to whether entry-level postings are growing, not just senior ones.
Match Your Strengths to a 4-Out-of-4 Major →
Our quiz scores 32+ majors against this framework using current BLS and AI-exposure data, then matches them to how you actually work.
Worked examples
Example A: Computer science
- Q1 (Licensed/scarce): Partial — no license, but senior CS is scarcity-protected.
- Q2 (Physical/trust/judgment): Yes for systems, architecture, security roles.
- Q3 (Depth that AI amplifies): Strong yes.
- Q4 (BLS growth): Yes — +17% through 2034.
Verdict: Strong major. Entry-level is harder than it was; specialize early.
Example B: Nursing (BSN)
- Q1: Strong yes — RN license required.
- Q2: Strong yes — physical care, human trust.
- Q3: Yes — AI clinical decision support amplifies expertise.
- Q4: Yes — +6% growth, severe shortage in 2026.
Verdict: Among the strongest majors in the AI age.
Example C: Generic business administration
- Q1: No.
- Q2: Partial — depends on specialization.
- Q3: Weak — credential doesn't add depth AI lacks.
- Q4: Mixed — +5% growth headline but heavy entry-level compression.
Verdict: Weaker as a generic credential. Strong if specialized into operations, finance, supply chain, or HR analytics.
Example D: Civil engineering
- Q1: Yes — PE license track.
- Q2: Yes — physical infrastructure work.
- Q3: Yes — AI-assisted design and BIM amplify the engineer's output.
- Q4: Yes — infrastructure investment driving demand.
Verdict: Strong major. Often underrated.
Example E: Graphic design (general)
- Q1: No.
- Q2: Partial.
- Q3: Mixed — generative AI competes with the credential.
- Q4: -4% growth in BLS projections.
Verdict: Weaker as a generic path. Pivot to UX, brand strategy, or specialty design (motion, 3D).
Notes for parents
If you're a parent helping a teenager think this through, three practical points:
First — don't optimize for the headline. The careers your high-schooler heard about on TikTok last year may have already changed. Use the four-question framework against actual 2026 data, not vibes.
Second — the safest answer is rarely the most exciting. Nursing, engineering, accounting (CPA-track), the trades, and special education are all in the strongest tier. They're not glamorous. They are durable.
Third — fit still matters most. A "safe" major your teenager hates will end in switching, dropping out, or grinding through a job they resent. Match the major to who they actually are, then optimize within that for AI durability.
Help Your Teen Find a Major That Fits Them →
Our quiz blends personality fit with 2026 labor data. It's built for parents and students who want to make this decision well — not panic into it.
Frequently Asked Questions
What's the single best major for the AI age?
There isn't one. The best major is the one that scores well on the four-question framework and matches how the student thinks. Nursing, engineering disciplines, computer science (with depth), and accounting (CPA-track) all clear all four bars for the right student.
Are liberal arts majors a bad bet now?
It depends. Generic English or generic communications without specialization is weakened. But philosophy, history, or English plus law school, plus journalism specialization, plus a teaching credential, or plus a complementary minor (data analytics, computer science) can lead to strong outcomes.
Should I just go to trade school instead?
For some students — especially hands-on learners — yes. The trades clear all four framework questions cleanly, costs are lower than a 4-year degree, and median wages are rising in 2026. See our trade school vs. college breakdown.
What if I'm already in a "weaker" major?
You don't need to switch. You need to specialize. A communications major can build a strong path in PR strategy, executive communications, or content strategy. A business administration major can build a strong path in operations or finance. Specialization is the fix.
How often should I revisit this framework?
Every 12-18 months. The labor data and AI capabilities are still moving fast. The framework holds, but the answers can shift.
Sources
- Stanford HAI. "AI Index Report 2025."
- McKinsey Global Institute. "Generative AI and the future of work in America," 2023, 2025 update.
- Dell'Acqua et al. "Navigating the Jagged Technological Frontier." Harvard Business School, 2023.
- Bureau of Labor Statistics. Employment Projections 2024-2034. bls.gov/emp.
- Bureau of Labor Statistics. Occupational Employment and Wage Statistics, May 2024.
- Brookings Institution. "AI Exposure Index" (Muro, Whiton, Maxim).
- Acemoglu & Restrepo. "Tasks, Automation, and the Rise in U.S. Wage Inequality." NBER 2025 update.