What College Is For When the Degree Stops Paying
If we can only defend the value of college as a salary, we've already lost the argument. The best case for college was never a paycheck.
Word crashed at midnight and took thousands of my annotations with it. I was a year into an education doctorate at Johns Hopkins, reading twenty to thirty research articles a week (averaging 300-500 pages). That evening of the crash, I’d been scrolling back through hundreds of pages of my own marginal notes, hunting for a note about close reading strategies that I knew I had written somewhere in the margins. The file reopened, but it was a version from a few months before. I sat there and did the arithmetic: ten to fifteen hours a week of reading, kept up for a year with several months of work lost, and no reliable way to find anything I had already written down. I cried.
Nobody had ever taught me to read research articles critically, much less systematically. I had a master’s degree, five years of science teaching experience, more years of science journalism bylines, and a drawer full of clips from magazines that pay by the word. Still, I couldn’t explain how to critically take a research paper apart.
Fifteen years before that, I was filing for Nature and Science on five-day turnarounds. My job was to read a study I really didn’t understand, call some people who did, and come back with the information that mattered for the story I was reporting. I learned this skill on deadline, from editors who bled red ink all over my drafts and then gave me grace to revise. I kept the marked-up pages in a folder for years. Nobody called that instruction, but it most certainly was.
Later, I taught high school lab science in San Diego, about 10 miles from the US-Mexico border with resettled refugee students from all over the world. Every fall, I watched the same thing happen as school got underway and everyone had to take my gatekeeper chemistry course. My students asked the same two questions: what would be on the test, and what was any of it for? I gave them replies about college, jobs, and earnings, and I could see how little of what I said resonated with many of them. Fair question, though, and my students have plenty of company.
In an NBC News poll conducted October 24 to 28, 2025, 63% of 1,000 registered voters said a four-year degree was not worth the cost, against 33% who said it was. In 2013, the same question split 53 to 40 the other way. A Wall Street Journal-NORC survey had already put the figure at 56% in March 2023. Among voters without a degree, it now runs 71%.
So the salary pitch appears to be finished, and the institutions that spent a generation making it have almost nothing to say in its place. If the whole defense of four years and a small fortune was a better job and a bigger paycheck, then the defense was always a number, and numbers move. As educators, we conceded the terms ourselves, and we did it because earnings were the easy thing to measure; the critical thinking was not. College became a vending machine: insert tuition, receive a better salary. No wonder people are voting with their wallets. We taught them to see it that way.
But the older answer, before jobs and earnings, was that the point of the four years was learning to be hard to fool, to think critically, and that capacity has never been more needed and harder to buy.
In October 2025, 22 public service media organizations across 18 countries tested four AI assistants on 3,000 news answers in 14 languages. Forty-five percent of the responses contained at least one significant issue. Thirty-one percent had sourcing problems: attributions that were missing, misleading, or invented outright. One assistant failed on 76% of its answers. As Jean Philip De Tender, media director and deputy director general of the European Broadcasting Union, explained when the study was released, the failures were not isolated incidents but systemic, cross-border and multilingual.
My high school students weren’t asking a lazy question. They were asking a serious question that matters even more now with frontier AI models advancing rapidly every few months. I no longer believe the honest answer to it is a salary figure.
Forty to sixty percent of American doctoral candidates finish every requirement except the dissertation, and that rate has held since the 1970s. They’re not lazy either, and every one of them was admitted on the assumption that somebody further back had taught them to critically read the scientific research literature. Mostly nobody did. As one doctoral student I interviewed put it, journal club is where “everyone just sits there and hopes that they don’t get called on with a question or ask questions.” The skill is real, it is teachable, and it is almost never written down on a syllabus. That’s what a hidden curriculum is: the thing students are graded on, need to succeed, and never taught.
An AI will now produce a fluent paragraph on any subject in about four seconds, and it cannot reliably tell us if the paragraph is accurate. Neither can a credential when the thinking is offloaded to an AI, but the grade still assumes the thinking happened. Who I want to hire and work with is the person who has sat with a hard text long enough to find the place where its argument fails. That is slow, and it costs time and mental energy. There’s no genuine version of it that I know about that arrives in four seconds.
Read deeply,
Dr. Genevive