What's Actually Happening to Entry-Level Design Jobs Right Now

I used to keep a folder on my desktop called “Junior Reqs.” Every entry-level design or UX researcher role I recruited for went in there: new-grad resumes, portfolio links to school projects, a cover letter or two about a cat rescue app. I could tell within ninety seconds whether someone had potential. That was the job.

A senior person with an AI tool could cover more ground than two juniors who still needed six months of ramp-up, and I understood that math because I’d been quietly doing a version of it myself already. I closed the reqs and told the candidates in final rounds the role was “on hold.” That phrase does a lot of quiet work in recruiting.

ServiceNow CEO Bill McDermott recently speculated AI could push the new-grad jobless rate as high as 30%. That’s one executive’s guess, made in public, not a data-backed forecast, and I don’t treat it as a stat. But it’s the kind of quote that spreads because people already feel like something’s wrong, and in the last year several independent research groups actually went and measured whether that feeling holds up.

I already made the case that “AI is coming for everyone’s job” is too simple a story in my post on whether AI is going to take our jobs. That argument still holds for the workforce overall. This post is the rung down from it, because when you isolate entry-level and junior roles in AI-exposed fields like design and software, the story stops being reassuring.

This is measured now, not just felt

The unemployment reversal is the headline most people haven’t seen stated plainly. The New York Fed’s own research on the labor market for recent college graduates found unemployment for grads age 22-27 hit roughly 5.7% in Q1 2026, against about 4.2% for the workforce overall. Recent grads have historically had a lower unemployment rate than the general population, not a higher one. This is the first time in the modern data that trade has flipped. Underemployment, grads working jobs that don’t require their degree at all, sat at 41.5% that same quarter.

The Stanford Digital Economy Lab, led by Erik Brynjolfsson, pulled ADP payroll data across millions of workers and found employment for 22-25 year olds in the most AI-exposed occupations, software development and customer service, has fallen roughly 16% since late 2022. That decline has since sharpened to about 3.8% a year, while employment for older, more experienced workers in the same occupations stayed flat or kept growing. Same job title, same company in a lot of cases, opposite trend line depending on how many years you’ve been doing it.

A Harvard-affiliated working paper looked at resume and job-posting data across roughly 280,000 U.S. firms from 2015 to 2025 and found something structurally similar: at firms that adopted generative AI, entry-level employment fell about 9% within six quarters relative to firms that hadn’t, while senior employment at those same firms kept growing. You may see this study cited elsewhere with an “80% drop” headline. That’s press framing of the same dataset, not a separately verified number. The defensible figure, and the one I’m using, is the 9%-within-six-quarters one.

Revelio Labs found entry-level job postings down roughly 35% since January 2023, more than 100,000 fewer postings a month, with tech-sector entry-level demand down about 25%, the steepest drop of any sector they tracked. Search volume for terms like “junior developer jobs,” “junior UX designer jobs,” “entry level UX designer jobs,” and “entry level tech jobs” has itself declined 33% to 65% year over year. On its own, that’s circumstantial. Lined up next to the data above it, it stops looking like a coincidence.

Stack those sources next to each other and you get something research rarely hands you this cleanly: independent teams, using entirely different methods, payroll records, resume data, official unemployment stats, job-posting counts, all landing on the same shape. Entry-level is shrinking faster than the market around it, concentrated in exactly the AI-exposed fields this site’s readers work in. If you want to sit with these numbers past this post, the design job market dashboard tracks the wage and posting trends behind this shift. It doesn’t yet break the data out by experience level, though. The report just ahead does exactly that, nationally. That’s a gap worth closing, and I’d bet it gets closed soon.

The same shift shows up somewhere you wouldn’t think to check: Google Trends. U.S. search interest in “will ai replace humans” is up more than 5,000% over the past five years, a jump big enough that Google’s own tool flags it as a breakout. The line barely moves through 2021 and most of 2022, spikes and drops twice as ChatGPT-era news cycles hit, then climbs almost continuously from mid-2023 into a peak in early 2026.

The related searches underneath that spike are the interesting part. “ai will not replace humans” tops the list, ahead of “what is ai” and “will ai replace jobs,” and it outranks “chatgpt” by more than ten to one. The fastest-growing term isn’t the reassurance-shaped one, though. “what jobs will ai replace” is up 250% even though it ranks lower, and “will robots replace humans” is the only term on the list actually declining, down 20%, like the whole question swapped “robot” for “AI” and kept the anxiety.

None of that proves anything about hiring by itself. Search behavior isn’t a labor statistic. But it’s the same shape as everything above it: quiet, then measured, then too big to wave off as a mood.

That anxiety has a clear source. Anthropic’s Dario Amodei told Axios in May 2025 that AI could wipe out half of all entry-level white-collar jobs within one to five years and push unemployment to 10 to 20 percent, warning the industry needed to stop “sugarcoating” what he called a coming white-collar bloodbath. OpenAI’s Sam Altman spent roughly the same stretch talking about the same disruption in softer language, but the substance matched. For the better part of two years, the men running the two companies whose models everyone’s asking Google about were also the ones setting the doom narrative.

In May 2026, within the same month, both walked it back. Altman told a Sydney banking conference he was “delighted to be wrong,” that the jobs apocalypse he’d warned about hadn’t materialized, and that he’d expected more entry-level damage by now than had actually happened. Amodei reframed his own prediction around the same time, describing AI as a multiplier that expands the 10 percent of a job you still do until it covers the other 90, not a destroyer that erases it. Same walk-back, same month, right as both companies were courting IPO investors who reward growth stories more than doom ones.

None of that makes the Stanford or NY Fed numbers above less real. It just means the two loudest, most credentialed voices on this topic spent two years pointing at a cliff that, by their own admission, wasn’t where they said it was.

The twist most coverage misses

Entry-level postings haven’t vanished. They’ve just stopped being entry-level. Indeed Hiring Lab published two companion reports on July 23, 2026, and the finding that stood out most wasn’t about AI directly. It was about who’s applying. In May 2026, 62% of applications from job seekers with 6-9 years of experience went to entry-level postings, versus 32% to mid-level roles and 6% to senior ones. Workers with 10 or more years of experience sent nearly half their applications to entry-level postings too. That’s the applicant list I noticed on LinkedIn Recruiter, confirmed at national scale.

This reframes the squeeze. It isn’t only AI erasing junior headcount from the demand side. Laid-off, experienced workers are moving down-market and applying directly for the roles that used to be new-grad on-ramps, squeezing supply from the other direction at the same time. A new grad isn’t just competing against fewer open reqs. They’re competing against candidates with a decade of shipped work applying for the same title.

That squeeze points to something none of the datasets above measure directly: what happens next. Fewer entry-level hires today means fewer people spending the next five years becoming a senior, and you can’t shortcut that experience no matter how good the AI tool is. The senior side of the ladder isn’t the safe landing it used to be either. Layoffs have hit experienced workers too, which is part of why so many of them are the ones filling out entry-level applications instead of senior ones. Some coverage is already calling this what it is: the entry-level squeeze isn’t just a hiring problem today. It’s a pipeline problem for who’s senior five years from now.

I understand the hiring-manager math here better than most people writing about it, because I used to build the business case for it. An argument circulating right now, bluntly stated in “AI didn’t kill your junior pipeline. You did,” says the real driver isn’t AI capability. It’s cost. A senior designer or engineer with an AI tool costs less than three juniors and produces output faster in the short term. I made a version of that argument to hiring managers before generative AI was even the tool in question, back when the trade-off was one senior versus two mid-levels. AI made the math starker. I didn’t invent the spreadsheet, but I used it.

Here’s the wrinkle that keeps this from being pure doom: the New York Fed’s own researchers argue roughly 64% of the recent rise in young-grad unemployment traces back to shifts in remote-work applicant pools, not AI directly. Remote postings pull in a wider, more experienced applicant pool than a local, in-office req ever did. That’s a labor-market structure problem layered on top of the AI story, not proof AI alone did this, and it changes what you should actually do about it, which is what the tactics in how to apply for jobs in a tough market are built around.

Is UX design dying, or just harder to break into

Design isn’t a special case in this data, but it isn’t a lost cause either. Everything above applies to design because UX and product design sit inside the same AI-exposed category as software development in the Stanford research: knowledge work with a heavy “produce a deliverable” component. Wireframes, first-pass copy, competitive audits, rough prototypes: a lot of what used to be a junior designer’s on-ramp task list is exactly what AI tools do a passable job of today.

BLS’s own occupational data shows that exposure landed unevenly. Between 2022 and 2025, the years spanning ChatGPT’s move from launch to mainstream workplace tool, entry-level (age 20–24) employment held roughly flat across all occupations and among software developers specifically. Entry-level graphic designer employment was cut nearly in half over the same three years.

Zoom out to the major occupational groups those two roles sit inside and the collapse looks contained rather than field-wide: the computer occupations group, which is where BLS actually files Web and Digital Interface Designers, its formal name for most UX and product design roles, grew slightly at the entry level, while the broader arts, design, and media group that graphic designers belong to slipped only modestly. Graphic design’s drop is concentrated in that one occupation, not the group around it.

That’s what’s behind the quiet rise in people searching “is UX design dying.” It’s still a small term in absolute terms, but it’s growing, which tells you something the bigger, declining “will AI replace designers” searches don’t: people aren’t asking whether AI is a hypothetical threat anymore. They’re asking whether the field is already over, present tense. I don’t think it is. I think the entry point moved and got narrower, a different problem with a different fix.

Zoom out from the entry-level squeeze to the occupation itself and that’s not a dying field. BLS’s Web and Digital Interface Designers category, its formal name for most UX, UI, and product design roles, is a mid-six-figure-employed occupation with a decade of above-average growth ahead of it.

For balance: not everyone reads this data the same way. Forbes contributor Hessie Jones has argued AI is not killing entry-level jobs so much as reshaping which entry-level skills matter. I don’t think that fully explains a 16% employment drop or a historic unemployment reversal, but the underlying point, that the skills getting hired for have shifted and not just the headcount, is worth holding onto.

What this means tactically: your portfolio can’t just prove you can execute a clean Figma file, because that’s the exact task AI is fastest at doing badly-but-passably. It needs to prove judgment: why you made a call, what you tried that didn’t work, what trade-off you accepted and why. That’s what a strong case study presentation is built to show a panel, and it’s why building your first portfolio around your judgment, not your polish, matters more now than it did five years ago.

That growth isn’t spread evenly across the map. BLS breaks Web and Digital Interface Designers down by metro area too, and the concentration is stark enough to matter for where you’re looking.

Reflective Coda

If you’re early in your design career, the honest thing to tell you is that you didn’t do anything wrong, and the advice that worked for the designer five years ahead of you is partially out of date. The entry point narrowed. It didn’t close. Companies still need people who can think clearly about a user’s problem, defend a decision in a room, and catch what an AI-generated first draft gets wrong. Those skills don’t show up in a “produce a deliverable” prompt, and they’re what a hiring panel is actually screening for now, whether or not they’d phrase it that way.

Treat this data as a map, not a verdict. You’re competing against more experienced applicants for the same postings, so your case for being hired has to be about judgment and fit, not just being cheaper and less senior. The search itself will likely take longer than you were told to expect, and that’s a market condition, not a reflection of you. If that weight is already sitting on your chest, that exhaustion has a name and it’s not a personal failing.

I still think about that folder of junior reqs, and about the candidates I told were “on hold” instead of the truth. I don’t recruit anymore, so I can’t fix that math from the inside. But I can tell you what I’d tell any of them today: the door got narrower, not shut. Bring something to the table a spreadsheet can’t replace with a senior hire and an AI subscription, and you’re not applying for the job that disappeared. You’re applying for the one that’s still worth having.

Sources

• “AI Is Already Taking Jobs Away From Entry-Level Workers” (Ina Fried, Axios, Aug 2025)
• “‘It’s Not Going Away’: The Stanford Economist Who Called the AI Entry-Level Jobs Crisis Early Has the Receipts” (Fortune, Jun 2026)
• “A New Stanford Analysis Reveals Who’s Losing Jobs to AI” (TIME)
• “Generative AI as Seniority-Biased Technological Change: Evidence from U.S. Résumé and Job Posting Data” (Seyed M. Hosseini & Guy Lichtinger, Harvard University, Oct 2025)
• “Why Entry-Level Hiring Is Down 80% At Companies Adopting AI” (Caroline Castrillon, Forbes, May 2026)
• “Labor Market for Recent College Graduates” (Federal Reserve Bank of New York)
• “Unemployment for Recent College Grads Remains High, NY Fed Says” (Bloomberg, May 2026)
• “Is AI Responsible for the Rise in Entry-Level Unemployment?” (Revelio Labs)
• “AI Isn’t Just Ending Entry-Level Jobs. It’s Ending the Career Ladder” (CNBC, Sep 2025)
• “Why AI May Kill Career Advancement for Many Young Workers” (CNBC, Nov 2025)
• “The Labor Market Is Tilting Toward Seniority” (Indeed Hiring Lab, Jul 2026)
• “Entry-Level Jobs Aren’t Just for Inexperienced Workers” (Indeed Hiring Lab, Jul 2026)
• “Employed Persons by Detailed Occupation and Age” (U.S. Bureau of Labor Statistics, Current Population Survey, Table 11b, 2022–2025 annual averages)
• “Web and Digital Interface Designers” (U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, SOC 15-1255, 2025)
• “AI Is Not Killing Entry Level Jobs” (Hessie Jones, Forbes, Nov 2025)
• “AI Didn’t Kill Your Junior Pipeline. You Did” (Exec Engineering, Substack)
• “will ai replace humans”: interest over time and related queries (Google Trends, United States, past 5 years, accessed Jul 2026)
• “AI Jobs Danger: Sleepwalking Into a White-Collar Bloodbath” (Axios, May 2025)
• “Sam Altman and Dario Amodei Are Both Walking Back AI Jobs Apocalypse Predictions as They Eye IPOs” (Fortune, May 2026)