October 10, 2026

How AI Is Changing What Majors Are Valuable

how ai is changing what majors are valuable

If you're picking a major right now, you've probably seen at least one headline telling you that AI is about to make your degree worthless. You've also probably seen the opposite headline, telling you to major in computer science because "AI is the future." Both of those takes are too simple, and the real data on what's happening to entry-level hiring right now is more interesting, and more useful, than either one.

Here's the short version: AI isn't wiping out majors. It's rewiring which tasks inside a career are still worth paying a 22-year-old to do. That distinction changes everything about how you should think about picking a major, and it explains some genuinely surprising numbers coming out of 2025 and 2026 research.

What the data actually shows

Let's start with the study that got the most attention: Stanford's Digital Economy Lab analyzed payroll data across millions of workers and found that employment for 22-to-25-year-olds in the most AI-exposed occupations (think customer service, accounting, and software development) has fallen noticeably since 2022, with the gap between young and experienced workers in those fields widening from 13% in an earlier read to 19% by mid-2026. Crucially, the researchers found this is happening mostly through reduced hiring, not layoffs. Companies aren't firing junior people in droves; they're just not opening as many junior slots in the first place.

Why would that be? The Stanford team's explanation is the single most useful idea in this whole conversation: employment is falling in jobs that rely on "codified knowledge" (the kind of stuff you can look up, template, or pattern-match, which AI is good at) and holding steady or rising in jobs that rely on "tacit knowledge" (the kind of judgment you only build through years of mentorship and hands-on repetition, which AI can't yet replicate).

The Burning Glass Institute, working with Harvard Business School's Managing the Future of Work project, reached a similar conclusion in its July 2025 report "No Country for Young Grads." Using Lightcast data covering nearly all US job postings, they found that the number of growth-track postings requiring under three years of experience has been falling since 2022 even as senior-level demand holds up. Unemployment and layoffs, they found, have risen fastest for 22-to-27-year-old degree holders in business, finance, and technology, the fields where a lot of junior work used to be exactly the kind of repeatable analysis, first-draft writing, and basic coding that generative AI now does in seconds.

Then there's the number that surprises almost everyone: according to the Federal Reserve Bank of New York's Labor Market Outcomes for College Graduates data, the unemployment rate for recent computer science graduates has climbed to around 6.1%, and computer engineering to roughly 7.5%, both well above the overall recent-graduate average and above fields like art history (about 3%) and nutritional sciences (well under 1%). That does not mean coding is a bad skill. It means the entry-level rung of that specific career ladder, junior developer roles built around writing routine code, has been the one most directly automated, while demand for experienced engineers who can architect systems and manage AI tools remains strong.

Why this is happening: codified vs. tacit skills

It helps to stop thinking in terms of "majors" and start thinking in terms of tasks. Every major bundles together a mix of codified tasks (research, first drafts, basic analysis, routine code, standard bookkeeping) and tacit, judgment-heavy tasks (persuading a client, diagnosing an ambiguous problem, managing a team through conflict, making a call with incomplete information). Anthropic's own Economic Index, which tracks how people actually use its Claude models, backs this up: coding and technical writing dominate current AI usage, and usage clusters most heavily around office and computational work rather than jobs built on physical presence or high-stakes interpersonal judgment.

The World Economic Forum's Future of Jobs Report 2025, based on a survey of over 1,000 employers representing more than 14 million workers, frames it the same way at a macro level: it projects 170 million new roles created and 92 million displaced by 2030 (a net gain of 78 million), while flagging that nearly 40% of core job skills will change in that window. AI and big data top its list of fastest-growing skills, but so do creative thinking, resilience, flexibility, and curiosity, which are exactly the human, tacit-knowledge skills that don't automate away.

The comparison: more exposed vs. less exposed

No major is purely "safe" or purely "at risk." But based on the reports above, some fields currently bundle more automatable, codified tasks than others. Here's how that's shaking out right now:

More exposed to AI disruption (entry-level) Less exposed / currently resilient
Junior software development, QA testing, routine coding Nursing and direct patient care (BLS projects healthcare adding roughly 2 million jobs this decade)
Entry-level accounting, bookkeeping, basic financial analysis Skilled trades and hands-on engineering roles requiring physical judgment
Paralegal document review, basic contract drafting Advanced practice healthcare (nurse practitioners: ~40% projected growth)
Customer service and call-center support Roles built on in-person negotiation, sales relationships, and trust-building
Entry-level marketing copywriting, first-draft content Skilled management and mentorship-heavy senior roles across all fields
Basic data entry and routine research summarization Fields emphasizing critical thinking, ethics, and communication (business, humanities, social sciences graduates applying these skills directly)

A caveat worth sitting with: this table describes task exposure, not major-level doom. A computer science major who becomes fluent in AI tooling, systems design, and infrastructure is generally landing well; it's the generalist junior coder doing purely repetitive work who's squeezed. Same logic applies across the board.

The common misconception

The biggest misconception floating around right now is that AI makes entire majors, especially humanities and liberal arts, obsolete, while only computer science or engineering are "AI-proof." The data says almost the opposite on both counts.

On one hand, computer science and computer engineering currently have some of the highest recent-graduate unemployment rates in the New York Fed's dataset, higher than art history, English, or performing arts. On the other hand, the AAC&U's 2025 employer survey of over 1,000 executives and hiring managers found that 93% rate written and oral communication, critical thinking, and ethical judgment as important skills in recent graduates, and BlackRock CEO Larry Fink has been widely quoted making the same point about hiring priorities:

"We have more and more conviction that we need people who majored in history, in English, and things that have nothing to do with finance or technology."

That's not an argument that coding doesn't matter, or that you should ignore the job market and follow pure passion with no plan. It's an argument that "safe major" was always the wrong frame. The truth is more specific: within every major, the tasks that are easiest to codify are getting automated fastest, and the people who combine technical fluency with judgment, communication, and domain expertise are doing best, regardless of what their diploma says.

What employers are actually asking for

Two of the most credible, ongoing employer surveys back this up. NACE's Job Outlook 2025, surveying employer members and non-members alike, found nearly 90% of employers want evidence of problem-solving ability, around 80% want strong teamwork, and roughly 70%+ still want solid written communication, initiative, and technical skills, with close to two-thirds now using skills-based (not just degree-based) hiring for entry-level roles.

Meanwhile, LinkedIn's 2025 workplace data (covered by Forbes) shows that "AI literacy," meaning the ability to use AI tools effectively, evaluate their output critically, and know when human judgment should override them, is one of the fastest-growing skills employers list, and most employers aren't looking for AI engineers specifically; they want professionals in every field who can work alongside AI competently.

Put those together and a pattern emerges: employers want subject-matter competence, a foundational level of AI fluency (regardless of major), and the human skills that AI can't fake. That's a very different message than "only STEM survives."

So how should you actually pick a major?

A few practical takeaways, none of which require predicting the future perfectly:

  1. Pick a field you can go deep in, not just declare. Whatever you choose, whether it's nursing, marketing, biology, or computer science, the goal is to move past the codified basics into the judgment-heavy, experience-built parts of that field as fast as you can, through internships, research, and real projects.
  2. Add AI fluency on top of your major, don't substitute it for one. Learning to use AI tools well is now closer to a basic literacy than a specialization. It's additive to every major, not a replacement for choosing one.
  3. Don't dismiss healthcare, skilled trades, and other hands-on fields. BLS projects healthcare adding roughly 2 million jobs this decade, and advanced practice roles like nurse practitioner are projected to grow around 40% through 2034, among the fastest of any occupation tracked.
  4. Take communication and critical-thinking coursework seriously even inside a technical major. The AAC&U and NACE data both point the same direction: these aren't soft extras, employers are treating them as core signals of long-term value.
  5. Look at internships and hands-on experience as insurance. Since the disruption is concentrated in junior, codified tasks, the students who show up already able to do judgment-heavy work (through co-ops, research, and real client or lab experience) are the ones skipping the worst of the entry-level squeeze.

Bottom line

AI is genuinely changing the entry-level job market, and the data on that is real, not hype: junior hiring is down in AI-exposed roles like accounting, junior coding, and customer service, and that shift shows up clearly in unemployment data for recent graduates. But it is not eliminating majors wholesale, and it is not true that only computer science or engineering majors are safe; in fact, some of those fields currently show higher recent-grad unemployment than humanities majors. What actually predicts good outcomes is depth within your field, real-world experience that builds judgment early, basic AI fluency layered on top of whatever you study, and the communication and critical-thinking skills employers keep saying they can't find enough of. Choose a major you can commit to and get genuinely good at, not one you're picking purely out of fear of a headline.

FAQ

Is it true that computer science is no longer a good major because of AI? No, but the entry-level tier of that field has gotten more competitive. Recent computer science graduates have an unemployment rate around 6.1% per NY Fed data, higher than several humanities fields, largely because junior coding tasks are among the most AI-automatable. Long-term BLS projections still show software developer employment growing about 15% from 2024 to 2034, well above average, so the field remains strong for those who build deeper skills beyond routine coding.

Does this mean I should avoid majors like English or history? Not based on current data. Fields like art history, English, and nutritional sciences currently show some of the lowest recent-graduate unemployment rates in the New York Fed's dataset, and employer surveys from AAC&U and NACE consistently rank communication, critical thinking, and ethical judgment, skills these majors emphasize, among the most in-demand traits for new hires.

What is "AI exposure" and how do researchers measure it? It generally refers to how much of an occupation's day-to-day tasks overlap with what current AI tools can do well, usually codified, pattern-based work like drafting, basic analysis, and routine coding. Stanford's Digital Economy Lab and Anthropic's Economic Index both use versions of this approach, based on real usage data and occupational task breakdowns, to see where AI use and employment shifts are concentrated.

Are entry-level jobs disappearing across the board? Not across the board, but they are shrinking in specific white-collar categories. The Burning Glass Institute's "No Country for Young Grads" report found postings requiring under three years of experience have declined since 2022 in business, finance, and tech-adjacent fields specifically, while senior-level demand has held steady, and healthcare and skilled hands-on fields have kept adding jobs.

Should I just pick whatever major has the highest current employment rate? That's risky, because these numbers shift as AI adoption changes and as more students react to headlines and crowd into "safe" fields. It's more durable to pick something you can genuinely go deep in, pair it with real experience and basic AI fluency, and build the communication and judgment skills that employer surveys consistently say matter, regardless of which major you chose.

Is AI going to keep getting worse for new graduates, or could this turn around? It's genuinely uncertain, and researchers are careful not to overclaim. Stanford's team notes the effect has been widening over time (from 13% to 19% in their tracked data) rather than stabilizing, but they also find no evidence of widespread displacement of experienced workers, and the World Economic Forum's 2025 report still projects a net gain of 78 million jobs globally by 2030 alongside the disruption. The realistic read is that the entry-level ladder is being reshaped, not removed, which is exactly why building tacit, judgment-based skills early matters more than it used to.

Sources

Related Articles

common sat math mistakes to avoid

Common SAT Math Mistakes to Avoid on the Digital SAT

The digital SAT math section trips up students in predictable ways. Here are the most common mistakes — from Desmos misuse to careless errors — and how to fix them.

 how to handle a bad professor strategies

How to Handle a Bad Professor: Strategies That Actually Work

Stuck with a disorganized, unfair, or hostile professor? Here's the real escalation path—from office hours to the ombudsperson to a formal grievance.