AI-exposed college graduates face 13% earnings drop, study finds

A study reported by Campus Reform links a 13% earnings drop to recent college graduates working in occupations the researchers classify as most exposed to artificial intelligence.

The figure, drawn from the outlet's coverage of an academic working paper, places an early numeric anchor on a question graduates and career counselors have asked since ChatGPT's public release in late 2022: which degrees still pay their tuition back, and how quickly.

A 13% earnings drop means a non-exposed comparator earning roughly $58,000 in the first post-graduation year translates to about $50,460 in the exposed function — a year-one gap of approximately $7,540, or about $20 per workday.

Multiplied across a four-year horizon at static salaries, the cumulative loss reaches approximately $30,000, before considering forgone raises, equity grants, or promotion trajectories. The dollar framing matters more than the percentage framing for a household weighing whether a six-figure student-debt load still pencils out.

The finding lands on a cohort already navigating a thin entry-level market, in which postings data tracked by Indeed Hiring Lab ran below pre-pandemic baselines across multiple 2024 quarters.

What counts as "AI-exposed"?

Labor-economics papers in this strand typically score occupations by the overlap between their task inventories and the capabilities of generative models, using taxonomies such as the Eloundou and colleagues 2023 GPTs are GPTs framework or earlier Felten-Raj-Seamans indices.

Office-support, marketing-associate, certain paralegal and analyst-track posts tend to score at the top; skilled trades, in-person healthcare, and equipment-intensive construction tend to score at the bottom.

Campus Reform's write-up does not specify which index the underlying study uses, nor does it publish the occupations roster, so a reader cannot yet tell whether the hit falls on a narrow band of majors or on a wide swath of entry-level white-collar work.

What does 13% mean for an offer letter?

For someone weighing two offers, the calculus is concrete. If the AI-exposed role pays a 13% premium to a non-exposed comparator, it still nets ahead. If the two offers land within 13% of each other, the non-exposed role compounds the advantage in dollar terms.

Career counselors advising on stretched student debt should weight the wage gap against the four-year repayment clock, not against the diploma.

What is the study not telling us?

Campus Reform's headline carries the finding but not the underlying methodology. The write-up does not state the sample size, the institution or working-paper series, the geography (US, UK, OECD), the cohort years, the controls for major, GPA, or institution selectivity, or whether the 13% reflects a one-time entry shock or a persistent trajectory.

A skeptical reader should treat the figure as a strong signal pending the dataset table economists expect before a number travels. The next revision will determine whether 13% becomes the working estimate for the Class of 2025 or a headline that ages badly.