A Census Bureau working paper published this month puts a number on something families have sensed for two years. Graduates of the majors most exposed to AI are getting hired less often, and they are paid less when they do get hired.
The paper is Graduating into Disruption, by Census economists Cody Orr, Lee C. Tucker, and Lawrence Warren. It tracks outcomes by major rather than by job title, which matters, because your student picks a major years before any employer sees them.
What the Paper Found
The researchers ranked majors by how much of the work they lead to can be done by AI, then compared graduates before and after ChatGPT arrived in late 2022, and found the outcomes began to split almost immediately. For the most exposed tenth of majors:
- Likelihood of employment in the first quarter after graduation fell five percentage points.
- Earnings in that first full quarter fell 13 percent.
- About half of the earnings drop came from graduates landing in lower-paying sectors like restaurants and retail. The other half came from lower pay inside the sectors that still hired them.
The authors say that earnings hit is comparable to graduating into a major recession. Warren told The Chronicle of Higher Education it is “in the same ballpark as the earnings effect of a Great Recession.” Unlike a recession, though, which hits every graduate at once, this shock is falling on new graduates in specific fields.
The gap narrows as graduates move further into their careers, but the paper says it stays substantial for the most exposed majors. The Chronicle reports that two years out, those graduates were still earning about 5 percent less than the comparison group.
This lines up with an earlier Census paper by Tucker, which found that early-career hiring in the most AI-exposed industries dropped right after ChatGPT, and employment for 22-to-24-year-olds in those industries fell 12 percent over the following ten quarters.
Which Majors Are Most Exposed
The paper says the most exposed tenth of majors, the group behind the five-point and 13 percent numbers, is dominated by computer science, information systems, and other majors tied to software development. If your student is heading toward CS, this data is about them.
The authors are also clear that above-average exposure reaches well beyond software. Inside Higher Ed’s summary of the paper lists accounting, journalism, and engineering among the most exposed majors, with nursing and education among the least exposed. So a student choosing accounting because it looks like a stable, practical alternative to CS is not stepping outside this trend. They are moving to a different part of it.
The exposure score is built from the work a major’s graduates typically go on to do, not from the subject itself. A field ranks high when a large share of the tasks in its entry-level jobs are ones AI can already perform. Exposure is also not the same as being replaced. A high score tells you the entry-level work is changing fastest there, which is why the effects in this paper are concentrated in the first job.
What This Does and Does Not Mean
The damage in this paper is concentrated at the start of a career. Fewer employers are hiring new graduates from these majors into entry-level roles, and the ones who do are paying them less. The paper does not show that the careers themselves are disappearing, and the authors say it is too early to know whether these graduates will carry the gap for five to ten years the way recession graduates often do.
It also does not tell your student to abandon a field. Chasing whatever major looks safest this year is how a lot of students ended up in the most exposed programs in the first place. For a decade, CS was the default answer for a student without a plan, so the degree said the same thing as thousands of others, and the entry-level tasks that used to justify hiring those graduates are the tasks AI now does. The students most at risk are the ones who chose the major because it looked safe rather than because they had something they wanted to do with it.
What Your Student Should Do With This
First, choose the major for what you want to do with it, not for the label. A student who wants to build medical devices and studies CS to get there is in a different position than a student who picked CS because it was supposed to pay. The same applies to accounting, journalism, and engineering: the student with a specific reason for the field has something to build on when entry-level hiring tightens.
Second, look at outcomes by program, not national averages. The data behind this paper comes from the Census Bureau’s Post-Secondary Employment Outcomes (PSEO) program, which publishes graduate earnings by major for participating colleges. Your student can see what graduates of a specific program at a specific school actually earn.
Third, plan to have the first job lined up before graduation. When entry-level hiring thins out, an internship, co-op, or research position that converts to an offer is worth more than it used to be, because the degree by itself carries less weight with employers than it did.
What This Means for the Application
Admissions readers follow this data too. Answering “why this major” with “because it’s in demand” already reads as empty, and it will read emptier by the time this year’s applicants graduate.
The student who can name the problem they want to work on, explain why this field is the way in, and show something they have already done about it is writing a different application. That has always been true, and it matters more now that the major itself signals less. Grades and scores get an application read, but a clear account of what the student wants to do and why is what gives the reader a reason to admit them.
