What Students Should Actually Learn in College in the Age of AI
Every parent I talk to is asking the same question: what should my kid major in if AI is going to do half the entry-level jobs anyway? It is the wrong question.
Computer science enrollment fell 8.1% in 2025–26, the steepest single-year decline of any field per the National Student Clearinghouse. Recent CS grads now face 6.1% unemployment, well above philosophy (3.2%) and art history (3.0%), per the Federal Reserve Bank of New York. McKinsey's 2025 workplace survey found 51% of organizations are reducing entry-level hiring because of AI.
Parents read those numbers and panic-pivot. Three years ago, half our families were certain their student would major in CS. Today, half are certain their student should not. Both positions assume the major is the variable that matters.
The variables that matter are two things colleges have under-taught for a decade and that AI cannot replicate at any price.
Working a room
The first is the ability to walk into a room of people and engage with them.
This sounds basic. It isn't. A generation has come up managing most of their relationships through screens, and the muscles of in-person communication have atrophied. Eye contact. Reading the energy of a room. Knowing when to listen and when to speak. Asking a question that makes someone want to keep talking. Remembering a name and using it later. These skills move a 24-year-old from interview to offer, and from offer to mentor. None of them appear on a transcript.
The NACE 2025 Job Outlook survey tells the same story in employer language. Roughly 90% of employers screen for problem-solving, 80% for teamwork, and more than 75% for communication. Fewer than 40% screen by GPA. McKinsey's CEO recently said the firm is "looking more at liberal arts majors, whom we had deprioritized." Graduates who can sit across from a client and command attention are scarce, and scarcity sets the market.
Working a room is also the most practiceable skill on this list. Every social interaction is a rep. Office hours. A summer job. The dinner table. A networking event where the student knows no one. The college that builds this skill is the one where the student has to talk to people they did not know, on terms that are not their own.
Critical thinking
The second is critical thinking, and I will say it plainly: most students do not have it, and most colleges are not building it.
Critical thinking is a discipline. It is the habit of refusing to take a claim at face value, looking at it from more than one angle, weighing evidence against counter-evidence, and arriving at a defensible position. The way to build that muscle is the way it has always been built: read widely, think about what you read, discuss it with people who disagree with you, and let the conversation change your mind when the argument is better than yours.
This matters more in an AI-saturated world, not less. Generative AI now writes plausible-sounding answers to almost any question. The student who cannot evaluate whether the answer is correct will be wrong in expensive, consequential ways. Anthropic CEO Dario Amodei said in May, "If you automate 90% of the job, then everyone does the 10% of the job." The 10% is judgment. It is where a human decides whether to trust what the AI produced, refine it, or throw it out. A student who has never been asked to disagree with an author or defend a thesis under pressure cannot do that 10%.
Reading is the foundational habit. Not what the algorithm serves up. Books. Sources that conflict. Authors who reasoned through different problems in different centuries. Discussion is the second half. A student who reads alone develops opinions. A student who defends those opinions against someone smarter or more contrarian develops thinking.
This is not a humanities pitch. A CS student needs this as much as an English major; an engineer who cannot tell whether an AI-generated design is sound is not an engineer for long.
What this means for families this cycle
Pick a major the student can commit to seriously. Pick a school where they will have to talk to strangers, defend ideas in seminars, and read material that pushes them. The two skills above are the resume entries that survive contact with an AI-saturated hiring market.
Computer science is not dead. NACE's Winter 2026 Salary Survey shows CS starting salaries rose 6.9% to $81,535, the largest increase of any major. The students hired can do something AI cannot. The same will be true of every major within two cycles.
Stop chasing the safest major. Start building the most adaptable graduate.
Frequently Asked Questions
What skills should students focus on in college in the age of AI?
Interpersonal communication and critical thinking. Working a room and presenting under pressure are the skills employers rank highest in the NACE 2025 Job Outlook survey. Critical thinking, built by reading widely and discussing what you read with people who disagree with you, is the durable counterpart.
Is computer science still a good major in 2026?
CS remains strong for students who genuinely want it but is no longer a default safety choice. Recent CS grads face 6.1% unemployment, while CS starting salaries rose 6.9% to $81,535 in 2026. The students who succeed in CS can also communicate well and think critically about what AI produces.
How can students develop critical thinking in college?
Read widely, especially material that challenges your assumptions. Discuss what you read with people who disagree with you. Take seminar courses where you must defend a position under pressure. Avoid outsourcing thinking to AI.
Why does AI make soft skills more valuable, not less?
AI is automating routine cognitive tasks faster than relational ones. What remains is the work that requires judgment, persuasion, and trust. A graduate who can command a room is more valuable in an AI-augmented workplace, not less.
