If you're applying for entry-level jobs this year, the headline you keep hearing is that AI is cutting them. The more useful finding for your resume is that some of the employers still hiring juniors now expect more from them than a junior used to deliver.
Northeastern's survey of 8,174 people with a say in hiring is the source of both findings. Among respondents who said AI affected their decisions, 26% are hiring fewer entry-level candidates and only 7% are hiring more. Among employers still hiring juniors at a steady pace, 28% now expect more output or higher-level work from them.
So an entry-level resume in 2026 has to do more than describe a junior. It has to show someone already doing part of the next level's work.
What Northeastern's survey actually found about entry-level hiring
The survey found two separate effects. AI is shrinking junior hiring at some employers and raising the bar for junior hires at others. The 26% figure only covers the roughly half of respondents who said AI influenced their hiring. That works out to about 13% of everyone surveyed, so it is wrong to say "26% of employers."
Here is how the numbers break down:
- 8,174 hiring decision-makers responded.
- Roughly half said AI made a difference in their decisions.
- Within that half, 26% are hiring fewer entry-level candidates and 7% are hiring more.
- The rest reported steady hiring, and 28% of them expect more output or higher-level work from entry-level employees.
John Wihbey, a Northeastern professor and survey co-author, put it plainly: "The fears around entry-level jobs are real." They are. But they come from one group of employers, and your resume is mostly being read by the other.
Two groups of employers, two different problems
The employers cutting junior roles and the employers raising expectations are different pools of respondents. A resume won't change the first group's mind, because they have decided the work no longer needs a junior. The second group still has a junior seat to fill, but they now picture someone who can own a result rather than only finish assigned tasks.
That's why a list of coursework and "assisted with" bullets underperforms now. It describes the junior role these employers are moving away from. A bullet shaped like "I decided X, which led to outcome Y" describes the role they still plan to fill.
Are entry-level jobs disappearing because of AI?
In the most AI-exposed occupations, partly yes. Across the whole market, no. Employment of 22-to-25-year-olds is falling in the jobs most exposed to AI, while overall hiring for the Class of 2026 is still projected to grow.
The data points in different directions depending on where you look:
| Signal | Figure | Source |
|---|---|---|
| Workers aged 22 to 25 in the most AI-exposed jobs, Nov 2022 to Jun 2026 | Down about 11% | Stanford Digital Economy Lab |
| Same age group in the least AI-exposed jobs | Up about 10% | Stanford Digital Economy Lab |
| Senior-and-above titles, share of postings (June 2026) | 26% | Posting snapshot |
| Intern, junior, entry-level, new-grad and associate titles, share of postings | 9.0% | Posting snapshot |
| Projected hiring change for the Class of 2026 | +5.6% | NACE Spring 2026 |
| Employers discussing eliminating positions due to AI | 11% | NACE Spring 2026 |
The Stanford work (Chandar et al.) shows where the decline sits: among young workers in AI-exposed jobs, while the same age group grows elsewhere. The New York Fed found the hiring pullback was concentrated in jobs that require a college degree, which is the market most new grads are applying into. And in the June 2026 posting snapshot, senior titles outnumbered junior ones almost three to one.
Put that next to Northeastern's 28% and the picture is consistent. The junior work that survives is the layer above routine execution: choosing what to build, checking whether the output is right, and owning the result. Your resume needs to show you working at that layer.
How AI screening changes what an entry-level resume has to do
At more than half of employers, software reads your resume before a person does. 54% of hiring decision-makers use AI on resumes, and in tech that rises to 83%. A resume that doesn't use the posting's language and doesn't show outcomes is likely to rank below the ones that do.
The Northeastern authors write that "increasingly, software reads the résumé before a person does." The reason is volume. The number of applications companies receive has more than doubled since 2022. The filter got stricter because the pile got bigger.
That creates a trap. More applications lead to harsher filtering, which pushes candidates to send even more applications, and many of them send the same generic resume every time. If you have spent weeks sending one resume to every posting you find and hearing nothing back, the screener has probably ranked a version of you that was never written for those jobs. Sending more of the same thing makes it worse.
The fix is fewer, tailored applications. Tailoring means reordering and rewording your real experience so the outcomes this specific posting cares about come first, in its vocabulary. Doing that properly for every posting is slow, which is exactly the argument for sending fewer.
Only 1 in 37 entry-level profiles mention AI
In Refolk's index of professional profiles, 12,013 of 443,230 US entry-level Software Engineers, Data Analysts, Product Designers and Marketing Associates mention AI in their headline. That's about 2.7%. Meanwhile, about 40% of hiring decision-makers see AI use on writing or coding assignments as a plus.
12,013 of 443,230 Software Engineer, Data Analyst, Product Designer and Marketing Associate profiles in Refolk's index.
Here are the demand and the signal side by side:
| Metric | Figure | Source |
|---|---|---|
| US entry-level SWE, analyst, designer and marketing associate profiles | 443,230 | Refolk's index |
| Same profiles with "AI" in their headline | 12,013 (2.7%) | Refolk's index |
| Entry-level jobs employers say require AI skills | ~35% | NACE Spring 2026 |
| Entry-level job descriptions that list AI skills | 16.5% | NACE Spring 2026 |
| Hiring decision-makers who see AI coursework as a plus | ~40% | Northeastern 2026 |
Each row has a different denominator (profiles, jobs, postings, people), so treat this as a direction rather than exact math. The direction is still clear. Employers value AI fluency in junior hires far more than junior candidates show it.
In a 25-profile sample of the full group, current employers included Google, Figma, Microsoft, LinkedIn, Glean and Ashby. The AI-headline sample included Figma, Google, Meta and OpenAI. That's too small a sample to conclude anything, but it shows the AI-headline group isn't limited to AI startups.
Employers expect AI skills they don't write down
According to NACE's Job Outlook 2026 Spring Update, employers say more than a third of entry-level jobs require AI skills, but only 16.5% of entry-level job descriptions list them. That gap matters if your tailoring strategy is just mirroring the posting's keywords. You'll miss a requirement the hiring manager holds anyway.
The same NACE report found:
- 28% of employers want early-career hires who can use AI.
- Nearly 60% give interns projects that use AI tools.
Some employers have made it formal. Cosmax Group added AI proficiency as a hiring qualification. Most haven't updated their postings yet. So show AI use even when the posting doesn't ask for it.
How to write resume bullets that read like the next level
Write each bullet as a decision you made and the outcome it produced. Name AI as the method and name the check you ran on its output. That shape works for the employers expecting higher-level work from juniors and for the software ranking you first.
Every strong early-career bullet has four parts:
- Outcome. What changed, ideally with a number you can defend.
- Decision. The call you made that a task-doer wouldn't have made.
- Method. The tool or approach, AI included, named specifically.
- Verification. How you checked the result: tested, validated, caught, rewrote.
Here is the difference by role. Swap the bracketed parts for your real figures and the tools you actually used.
| Role | Task-list version | Junior-plus version |
|---|---|---|
| Engineer | Wrote unit tests for API endpoints | Chose to test the payment edge cases first; generated tests with an AI assistant, rejected [n] that asserted the wrong behavior, caught [n] bugs before release |
| Analyst | Created weekly dashboards | Rebuilt the weekly report to answer the churn question leadership kept asking; drafted the queries with Claude, validated them against finance's numbers, cut prep from [x] hours to [y] |
| Designer | Designed screens for onboarding flow | Cut onboarding from [x] steps to [y] after reviewing session recordings; explored layouts with AI, tested two with [n] users, shipped the one that won |
| Marketer | Wrote social posts and email copy | Ran an email subject-line test; drafted variants with AI, rewrote the off-brand ones, lifted open rate from [x]% to [y]% |
The right column is longer, and that's fine. It has the posting's skill words for the software and the evidence of judgment for the person who reads it next.
Frame AI as judgment, not volume
Show AI as a tool you checked, not one that did the work for you. 21% of hiring decision-makers say AI use on every assignment would discourage them from making an offer, and almost half said AI experience wasn't explicitly on their radar. "Used AI for everything" reads as a risk to them. "Used AI, then caught its mistake" reads as a skill.
Wihbey's argument points the same way. The critical thinking you build in college is what lets you write useful prompts and judge what AI produces. Put that judgment into your verbs:
- Caught: you found what the model got wrong.
- Validated: you checked output against a source of truth.
- Rewrote: you kept the structure and fixed the substance.
- Rejected: you threw out output that didn't meet the bar.
- Tested: you put the result in front of users or a test suite.
Nobody hires a junior for typing speed anymore. They hire the one who knows when the model is wrong.
New grad resume with no experience: what counts as owned work
Any project where you made a call and can point to a result counts. Coursework counts only once you rewrite it as a decision and an outcome. Since nearly 60% of employers already give interns AI-tool projects, an internship where you shipped something with AI is close to what the full-time role will look like.
Places to find owned work you may be underselling:
- Capstone and senior projects. Name the scoping decision, not just the deliverable.
- Internships. Even a short one usually has one thing you shipped end to end.
- Freelance or campus jobs. A club's sign-up flow you rebuilt counts as product work.
- Open-source contributions. A merged PR is an outcome with a public link.
- Self-directed builds. A tool you made to solve your own problem, with a user count if it has one.
Most new grads have plenty of this material. What they lack is the time to dig it out and phrase each piece for each posting. Refolk writes your resume from your own history for the specific posting, so it pulls these projects forward and frames them as outcomes without inventing anything. It also scores how well you actually fit each posting before you apply, so you can skip the jobs where the gap is real and spend your effort on the ones worth tailoring for.
What to cut to make room
Take out anything that describes the old junior role:
- Objective statements. The screener ranks on evidence, not intent.
- "Relevant coursework" lists. Course titles don't tell anyone what you decided.
- Bare tool lists. A row of tool names without a use attached is filler.
- "Familiar with" and "exposure to." Both signal that you haven't done the thing.
- "Assisted with" bullets. Rewrite them around the part you owned, or drop them.
FAQ
Are 26% of employers hiring fewer entry-level workers?
No. The 26% figure comes from the roughly half of Northeastern's 8,174 respondents who said AI affected their hiring decisions. As a share of everyone surveyed, it's about 13%. NACE's Spring 2026 update projects a 5.6% hiring increase for the Class of 2026, and only 11% of employers are discussing eliminating positions because of AI.
Should I put AI in my resume headline?
Yes, if you can attach it to a real result. In Refolk's index, only 12,013 of 443,230 US entry-level profiles in four common title groups mention AI in their headline, about 1 in 37. A specific phrase like "Data analyst who builds validated reporting with Claude" stands out to both the screener and the reader. A vague "AI enthusiast" does not.
Will mentioning AI hurt my application?
It can if you frame it as volume. 21% of hiring decision-makers would be discouraged by AI use on every assignment. Lead with the outcome, name AI as the method, and show your verification step. Framed that way, you're speaking to the roughly 40% who see AI-assisted work as a plus without worrying the people who don't.