DegreeWorth

The High-Ceiling vs. High-Floor Bet: How to Pick a Major When AI Is Eating Jobs

Everyone is asking the same question: which degrees survive AI?

The usual answer is a list. "Top 10 AI-proof majors." You scan it, see nursing at the top, feel vaguely reassured or vaguely worried, and move on. Those lists are useless because they flatten a three-dimensional problem into one dimension.

When we ranked all 287 college majors under an aggressive AI scenario, three distinct strategies emerged:

  1. High salary, low employment certainty — you're betting big on landing a field job
  2. High employment certainty, modest salary — you're betting on stability
  3. High salary and high employment — the degrees that actually have it all

Most people assume you have to pick between Strategy 1 and Strategy 2. You don't. Strategy 3 exists — and the degrees in it might surprise you.

24,479
Programs Analyzed
$542K
Avg 10yr EV
39%
Avg P(Field Job)
$888K
Best 10yr EV

How We Measured This

Every number in this article comes from one calculation: how much money will you probably make over 10 years if AI goes hard?

Expected Value = P(field employment) × field salary + P(underemployment) × fallback salary

Summed over 10 years with wage growth. Employment probability comes from Federal Reserve Bank of New York data, adjusted for AI disruption. Fallback salary is $45,000 (FRBNY median for underemployed grads).

Two numbers drive the result: how much you earn if you get a field job, and how likely you are to get one. Some degrees win on salary. Some win on probability. A few win on both. Here's what the data shows.

Strategy 1: Bet on Salary

High salary, but less than 50/50 odds of field employment

These degrees pay so well that even when AI cuts field employment below 50%, the expected value still dominates. You're betting you'll be in the fraction that lands a field job. If you miss, you fall back to $45K — not great, not catastrophic. This is a leveraged play: high variance, high expected payoff.

#Major10yr EVP(Field Job)Year 1 SalaryAI ExposureSchools
1Operations Research$888K46%$87,03554%6
2Petroleum Engineering$761K49%$64,10648%18
3Pharmacy$741K47%$53,14346%13
4Electrical Engineering$733K48%$77,51756%262
5Math & Computer Science$727K39%$92,84967%9
6Computer Engineering$727K43%$78,69571%174
7Computer Science$711K40%$73,71472%345
8Industrial Engineering$694K48%$73,87443%93
9Biomedical Engineering$688K48%$63,75150%119
10Systems Engineering$684K45%$75,19356%11
11Nuclear Engineering$682K49%$72,61149%9
12Materials Engineering$670K49%$71,34148%33

Ranked by 10-year expected earnings under aggressive AI. Filtered to P(field employment) < 50%. Minimum 5 schools.

Computer Science is the purest version of this bet. It has the highest AI exposure on this list (72%) and one of the lowest employment probabilities (40%). But the $74K starting salary is so high that even weighted by worse-than-coinflip odds, the EV beats almost every "safe" major.

The catch: that 40% is an average across 345 schools. At MIT or Stanford, your personal odds are much higher. At a mid-ranked regional school, probably lower. This strategy works best when you have every advantage: a top program, strong grades, and strong internships.

Strategy 2: Bet on Certainty

High employment probability, but modest salaries

These degrees have the highest probability of field employment even when AI goes hard — 55% or higher. The salary won't match a CS grad who beats the odds, but you're almost guaranteed to actually work in your field. This is the rational choice when you need reliable income to service debt, or when you don't have the luxury of gambling on a competitive job market.

#MajorP(Field Job)10yr EVYear 1 SalaryAI ExposureSchools
1Teaching (K-12)62%$475K$41,41230%679
2Social Work58%$487K$36,55524%338
3Special Education58%$491K$44,10544%170
4Human Services57%$455K$37,29226%84
5Urban Planning56%$592K$42,02335%15
6Subject Teaching55%$497K$41,69043%348

Ranked by P(field employment) under aggressive AI. Filtered to P ≥ 55% and 10yr EV below $690K. Minimum 5 schools.

The pattern: direct human contact. Teaching, social work, special education, human services — these survive because the work is irreducibly human. Someone has to be in the room with the student or the client. AI can write a lesson plan, but it can't manage a classroom of 8-year-olds.

The tradeoff is real. These degrees have 10-year EVs of $455–592K — roughly $200K less than the high-ceiling degrees. You're trading expected dollars for expected stability. For some people, that's a great deal.

Strategy 3: Best of Both Worlds

Here's where it gets interesting. A small group of degrees don't force you to choose between salary and employment. They rank in the top tier for expected earnings (above $690K) while also maintaining 50%+ field employment — meaning the salary is high and you're more likely than not to actually land a job in your field.

#Major10yr EVP(Field Job)Year 1 SalaryAI ExposureSchools
1Naval Architecture$791K51%$89,78245%5
2Aerospace Engineering$789K61%$73,06041%57
3Chemical Engineering$768K54%$72,28849%158
4Mining & Mineral Eng.$758K50%$84,28747%5
5Construction Engineering$732K55%$76,54348%14
6Nursing$716K64%$75,27439%990
7Construction Management$707K54%$72,73849%56
8Mechanical Engineering$704K50%$70,52753%320
9Civil Engineering$694K55%$69,09749%220

Degrees with 10yr EV ≥ $690K and P(field employment) ≥ 50% under aggressive AI. Both columns highlighted. Minimum 5 schools.

Nursing is the standout. It has the highest field employment on this list (64%), a $75K starting salary, and is offered at 990 schools. The FRBNY data shows only 12.8% of nursing grads end up underemployed — the lowest of any major they track. If you forced us to name a single best all-around bet in the dataset, this is it.

Aerospace engineering is the engineering version of the same story: 61% employment, $73K salary, $789K EV. You can't automate building rockets, and the people who build them get paid.

The common thread across the rest of the list: physical-world engineering. Chemical plants, construction sites, mine shafts, shipyards, bridges. These fields have enough real-world, hands-on work to resist AI displacement, while paying enough to generate EVs that rival the pure salary plays. They're not obscure niche degrees either — mechanical engineering alone is offered at 320 schools.

The Bottom of the List

What about the worst expected outcomes? These degrees have both low salary and low employment probability — the opposite of Strategy 3:

Major10yr EVP(Field Job)Year 1 SalarySchools
Dance$408K30%$22,51362
Drama & Theatre Arts$423K27%$21,147270
Culinary Arts$433K36%$28,61112
Pastoral Counseling$435K45%$31,97622
Visual & Performing Arts$440K30%$25,52559
Religion & Religious Studies$444K34%$26,70940
Music$445K24%$28,116240

Dance has only 24% AI exposure. Music is 47%. AI-resistance is not the same as a good bet. The performing arts are highly resistant to AI automation and still produce the worst expected earnings in the dataset, because the baseline employment rates were already bad before AI entered the picture.

So Which Strategy Is Right for You?

Strategy 1: Bet on Salary

• You're at or can get into a top-tier program (top 50 school for your major)
• You're competitive within that program (top third of class)
• You have a financial safety net (family support, low debt)
• You're drawn to CS, electrical engineering, applied math, or finance

Strategy 2: Bet on Certainty

• You need reliable income to service student debt
• You prefer stability over maximizing expected value
• You're drawn to teaching, social work, or human services
• Career satisfaction matters more than lifetime earnings to you

Strategy 3: Get Both

• You want strong earnings without betting against the odds
• You're willing to do physical-world, hands-on work (engineering, healthcare)
• Nursing, aerospace, mechanical, chemical, or civil engineering appeals to you
• You want the answer that doesn't require you to be in the top 10% to pay off

The trap to avoid: defaulting into a degree that's none of the three. Business administration ($521K EV, 32% employment) and psychology ($481K, 34%) have moderate salary, moderate-to-high AI exposure, and high underemployment. They're not high-ceiling, they're not high-floor, and they're definitely not both. A lot of students end up there by default.

See the Full Analysis for Your Major

Every program page on DegreeWorth now shows Probability of Field Employment under all three AI scenarios.

Browse All Majors

Methodology

This analysis covers 24,479 bachelor's degree programs at 1,879 schools with non-suppressed earnings data from the U.S. Department of Education College Scorecard.

Expected Value is calculated using an employment probability model:

EV = P(field employment) × W_target + (1 - P(field employment)) × W_fallback

Summed over 10 years with CAGR wage growth. In the pessimistic scenario, the AI severity parameter E = 0.7 (aggressive displacement), BLS growth offset is zeroed out, and displaced workers earn $0 in year one (frictional unemployment).

  • P(field employment) starts from Federal Reserve Bank of New York underemployment rates by major (~73 categories mapped to 287 CIP-4 codes), adjusted for AI disruption severity and BLS employment projections
  • W_target: College Scorecard median earnings per school + major, grown at observed CAGR (or 3% default when Year 5 data is suppressed)
  • W_fallback: $45,000 (FRBNY median wage for underemployed college graduates), grown at 2% inflation
  • AI exposure: weighted composite of OpenAI GPT Exposure (0.4), Felten AIOE (0.3), and Frey-Osborne automation probability (0.3)
  • BLS offset: Bureau of Labor Statistics 2024-2034 employment projections, capped at 90% of AI penalty

The pessimistic scenario is not a prediction. It models what happens if AI adoption is aggressive and rapid. Use it to stress-test your choices, not to predict the future.

Cite This Analysis

DegreeWorth. "The High-Ceiling vs. High-Floor Bet: How to Pick a Major When AI Is Eating Jobs." March 7, 2026. degreeworth.com/blog/high-ceiling-vs-high-floor-degrees.html