Gen Z Graduates In 'AI-Proof' Majors Are Earning 18% More Than Their Peers
Co-Founder and Chief Data Scientist
An 18% earnings advantage for Gen Z graduates in “AI-proof” majors is not reported by the two newest administrative studies of graduates and artificial intelligence. A U.S. Census Bureau working paper published in September 2026 found that initial full-quarter earnings for graduates in the most AI-exposed decile fell 13% relative to the six least-exposed deciles after ChatGPT’s November 2022 release.
On the Census paper’s rounded estimate, that shortfall makes the comparison group approximately 14.9% higher on the reciprocal calculation, still below 18%. More importantly, the finding measures a change relative to earlier earnings trends, not an absolute salary ranking among majors.
The Headline Measures A Change In Trajectory
The Census researchers analyzed 6,665,500 bachelor’s-degree records from 2012 through 2024, covering more than 350 institutions in 24 states and representing approximately 29% of U.S. bachelor’s degrees conferred from 2016 through 2024. The observational study compared changes within institutions and majors over time rather than simply comparing current salaries across subjects.
Exposure scores used occupations held by workers ages 21 to 29 with only a bachelor’s degree in the Census Bureau’s 2018 through 2022 American Community Survey. Those occupations were connected to a task-based measure of whether GPT-4 could reduce the time required for particular work activities by at least half.
Computer science, computer engineering, information sciences and programming dominated the most-exposed decile, while education, health, agriculture and several life-science fields ranked lower. However, the Census researchers noted that almost every major supplied some graduates to highly exposed occupations, and the scores measured technological capability rather than guaranteed adoption or job replacement.
The sample was not limited to Gen Z, and participating institutions were disproportionately public, large and research-oriented. The study is also a working paper that has not received the review applied to formal Census Bureau publications.
First Jobs Carry The Largest Penalty
The Census researchers estimated a 5-percentage-point relative decline in first-quarter employment probability for graduates in the most-exposed decile, alongside the 13% relative decline in initial full-quarter earnings.
Between the Census paper’s 2022 and 2024 cohorts, that group became more than 2 percentage points more likely to enter accommodation and food services and similarly more likely to enter retail trade. Its employment share fell by more than 3 percentage points in information and nearly 6 percentage points in professional, scientific and technical services.
According to the Census team, movement into lower-paying sectors accounted for approximately half of the earnings decline. The initial penalty diminished with time, although statistically significant earnings declines remained for the most-exposed graduates for at least two years, making the pattern more consistent with a damaged entry-level ladder than a permanent judgment on lifetime degree value.
Texas Confirms Direction, Not Size
A Federal Reserve Bank of Dallas analysis published on September 22, 2026 used administrative records from Texas four-year universities and found the same direction of change with a smaller estimated earnings effect.
For Texas graduates, a 10-percentage-point increase in tasks susceptible to generative AI automation was associated with a 1.7-percentage-point relative decline in employment likelihood within a year. Among graduates working in Texas, first-year earnings in more-exposed majors fell approximately 5% between the 2021 and 2024 classes relative to less-exposed majors.
The Dallas Fed also found that May 2024 graduates from majors with 10 percentage points more exposure were 1.4 percentage points more likely to re-enroll within a year. Two-thirds of returning computer science graduates entered another computer science program, yet new workers with master’s degrees in exposed fields experienced relative earnings declines similar to those with four-year degrees.
Lower Exposure Does Not Guarantee Higher Returns
The Federal Reserve Bank of New York’s college labor-market tracker reported approximately 5.6% unemployment and 42% underemployment among recent graduates ages 22 to 27 in the second quarter of 2026. Underemployment meant working in an occupation that typically does not require a college degree.
In the New York Fed’s April 2025 life-cycle return analysis, based on the Census Bureau’s 2023 American Community Survey and College Board and Education Department data, engineering, mathematics and computers, and business and economics produced the highest median returns. Education produced particularly low returns, while health sciences remained above average; the researchers cautioned that annual education earnings generally represent a nine-month school year.
Lower AI exposure can therefore accompany greater near-term stability without producing higher absolute earnings. An exposed technical degree can face a difficult entry market while retaining a stronger long-run wage premium.
Major Choices Require More Than Exposure Scores
The Census researchers found that the number of graduates in the most-exposed decile rose 24% from the 2017 academic year through 2022 while employment largely kept pace. From 2022 through 2024, annual graduate growth slowed to less than 3%, but the initially employed population declined 3%, making increased supply alone an incomplete explanation; the authors nevertheless acknowledged that their observational design could not eliminate every contemporaneous shock correlated with AI exposure.
The evidence supports a narrower conclusion than the headline: graduates from majors closely connected to software-intensive entry-level work have experienced substantial relative deterioration since late 2022, while less-exposed fields have so far been more resilient.
For students choosing a major, exposure should be considered alongside tuition, completion prospects, licensing or graduate-school requirements, absolute earnings and underemployment. A more stable first job can still pay less, while a degree facing greater disruption at entry can retain the stronger lifetime return.
