If you graduated in the past three years with a degree in computer science, accounting, journalism, or engineering, the job market waiting for you looks nothing like the one your professors described. A working paper by economists affiliated with the U.S. Census Bureau, titled Graduating into Disruption: Labor Market Outcomes for AI-Exposed College Majors, has now put hard numbers on what many recent grads already sensed: initial earnings for graduates in the most AI-exposed majors have fallen by 13 percent compared with the pre-ChatGPT era, and the probability of landing a first job has dropped by five percentage points. The authors frame that earnings hit as comparable in magnitude to graduating into a large recession, and that framing is worth holding onto as you read listings, weigh offers, and figure out whether to take the first paycheck in front of you.

What the research actually found

The researchers drew on administrative records covering college graduates across the country, then ranked majors into ten categories from most to least AI-exposed. Computer science, accounting, journalism, and engineering landed near the top of the exposure ladder. Nursing and education sat near the bottom. The comparison baseline is outcomes before large language models went mainstream, roughly four years ago, following ChatGPT's public release in late 2022.

The Census Bureau-affiliated team stated their central finding directly:

Post-graduation employment, earnings, and job switching patterns among the most AI-exposed college majors began to diverge immediately following the introduction of ChatGPT in late 2022. This earnings decline is comparable in magnitude to the earnings losses associated with graduating into a large recession.

Two numbers deserve your full attention. First, five percentage points fewer graduates from AI-exposed majors are getting hired at all during their initial job search. Second, those who do get hired are starting at salaries 13 percent below what similarly credentialed graduates in the same majors earned before generative AI became a mainstream workplace tool. Both figures come from comparisons against pre-2022 cohorts in identical fields, so this is not a story about a weaker overall job market. The Bureau of Labor Statistics (BLS) Employment Situation reports have shown continued low headline unemployment across the broader economy during this period, which makes the concentration of damage in specific majors even more striking.

Where the lost earnings are actually going

Here is the detail that changes what you should do about it. The pay decline is not one clean story about tech companies offering smaller starting packages. The Census Bureau-affiliated researchers split the gap almost in half:

Roughly half of the decline in earnings is attributable to lower earnings within the industry sectors that employ these graduates, with the remainder resulting from a shift in the industry mix into lower-wage sectors such as restaurants and retail. The effects attenuate as graduates move further from labor market entry but remain substantial for the most exposed majors.

Translated for your situation: about half of that 13 percent drop reflects graduates in AI-exposed fields accepting lower pay inside the industries they actually trained for. The other half reflects graduates from those same fields ending up in food service, retail, and other sectors that were never part of the plan. Both trends are real. Both carry different implications when you decide whether to hold out for a role in your field or take the available paycheck.

The researchers also note the effects ease as graduates move further from their initial entry point, but they remain substantial for the most exposed majors. In plain terms: the financial hit is worst at the first job, and it does fade over time, but not fast enough to treat as background noise while you're signing an offer letter.

What this means when you compare offers

If a starting salary feels low right now, the honest answer is: by pre-2022 standards for your major, it probably is. That is not a negotiating failure on your part. It reflects a market-wide reset that compensation professionals at WorldatWork have flagged in recent salary budget surveys as AI-driven compression at the entry level, particularly in technical and analytical roles.

A few things worth doing before you sign anything:

• Benchmark against current data, not old advice. Use a salary tool that reflects what people in your specific role and city are actually earning in 2026, and check employer reviews and posted pay ranges for companies you are considering. If your college career center is quoting graduate salary figures from a 2021 or 2022 survey, those numbers predate the shock and will lead you astray.
• Separate below-market from below-expectation. If an offer sits below the current market rate for your role, that is worth negotiating. If it is at current market but below what you hoped for two years ago, the negotiating lever has moved and you need updated data to push back effectively.
• Weigh the industry-drift risk carefully. A junior role in your field at 13 percent below the old benchmark is still, statistically, a stronger career trajectory than a stopgap position in a lower-wage sector. The Census Bureau-affiliated data indicates that graduates who slip into restaurants and retail account for roughly half of the total earnings gap, and the research suggests that drift is difficult to reverse quickly.
• Ask the employer directly about their approach to AI in the role. This is not a trap question. It is a fair and practical one. The answer tells you which parts of the position are being reshaped, where you would genuinely be adding value, and how stable the role is likely to be over a two-to-three-year horizon.

If you are still job-hunting: where to focus your applications

The five-percentage-point drop in initial employment is doing real damage to the volume side of the search. Fewer entry-level postings, more applicants per posting, and longer timelines between application and offer. The Society for Human Resource Management (SHRM) has noted a consistent increase in application-to-interview ratios for entry-level technical roles since 2023, reflecting both AI screening tools on the employer side and a larger pool of competing applicants. If you are in an AI-exposed major and still searching, a few shifts in approach can help.

Look at adjacent, lower-exposure sectors

A computer science graduate can write production software in a hospital network's IT department, a school district's data office, or a public-sector infrastructure team. An accounting graduate can join the finance function of a healthcare group or a logistics company. The Graduating into Disruption paper identifies these sectors as carrying lower AI exposure than the industries most CS and accounting graduates traditionally target. Choosing one of them is a hedge, not a retreat, and it keeps you in a role that builds transferable skills.

Make the parts of your work that AI cannot easily replicate visible on paper

On your resume and in interviews, a generic line about Python proficiency, financial modeling skills, or strong writing is doing less work than it did three years ago. A large language model can plausibly claim the same. What lands better is specific: a project outcome with a number attached, a client problem you solved, a bug you caught that automated tests missed, a story you reported that required building a source relationship over weeks. Concrete evidence of judgment outperforms keyword-optimized skill lists in a screening environment where AI tools are reviewing applications first.

Prepare a clear, specific answer about AI tools in your workflow

Expect every employer in an exposed field to ask some version of how you use AI tools in your day-to-day work. There is no single correct answer, but there is a clear wrong one: claiming you do not use these tools at all. Come ready with a specific example of what you use, what output you always verify, and what kinds of decisions you would never delegate to an automated system. That framing signals judgment, not technophobia or uncritical dependence.

Is this temporary or the new normal

The honest answer is that nobody knows yet, including the researchers themselves. They acknowledge they do not yet have sufficient data on how persistent these effects will prove to be. What they do conclude is worth reading carefully:

Even so, the evidence shows that highly exposed graduates have borne substantial costs in the initial adjustment to generative AI. Understanding whether these workers ultimately recover, and how the technology reshapes entry-level career paths and human capital accumulation, will be central to assessing the longer-run distributional consequences of AI.

Heidi Shierholz, President of the Economic Policy Institute (EPI), has argued in published EPI analysis that early-career wage disruption from technological displacement tends to have longer-lasting effects on lifetime earnings trajectories than disruption that hits mid-career workers, because the first few roles set the baseline for every subsequent raise and negotiation. That context makes the Census Bureau-affiliated findings more consequential, not less, even if the researchers themselves are appropriately cautious about long-run projections.

Some recent graduates in AI-exposed majors will catch up within a few years as the market adjusts. Others may spend the better part of a decade trying to reset a career that began in a sector that was never the intended destination. The Department of Labor (DOL) does not yet track AI-driven major-level displacement as a separate category in its occupational outlook data, which means the standard career guidance resources are not yet equipped to give you an accurate picture of this specific risk.

What you can control is the quality of your next two or three moves. Do not accept a below-market offer without first checking what current market actually means for your role and city in 2026. Do not drift into a stopgap sector without a clear timeline and plan for returning to your field. And do not rely on salary advice, graduate outcome data, or job search strategies that predate 2022, because the room you are negotiating in today is materially different from the one those resources were built to describe.

Key takeaways

• Recent graduates in the most AI-exposed majors, specifically computer science, accounting, journalism, and engineering, are entering the workforce with starting salaries 13 percent lower and initial employment rates five percentage points lower than pre-2022 cohorts in the same fields.
• The earnings gap splits roughly in half: lower pay within target industries, and a drift of graduates into lower-wage sectors like restaurants and retail.
• Benchmark any offer against current 2026 market data, not pre-shock figures from your college career center. A salary that feels disappointing may reflect the new market rate rather than a negotiating gap.
• In applications and interviews, lead with specific, verifiable work outcomes rather than generic skill claims that AI tools can also produce. Concrete evidence of judgment is the differentiator.
• The Census Bureau-affiliated researchers are candid that persistence of these effects is not yet established. Protect your first two career moves accordingly, because early-career wage baselines have long-run consequences for lifetime earnings.