Every article about the entry-level hiring collapse says roughly the same thing: AI is eliminating the roles that used to exist for new graduates. Prompts are replacing interns. Agents are handling the tasks that used to teach junior employees how to work. The logical conclusion is that the problem is structural and irreversible, a fundamental shift in what companies need from people.
New research from the Federal Reserve Bank of New York says the conventional wisdom is mostly wrong.
Published June 1, 2026, the NY Fed study examined unemployment among college graduates under 29 and traced what actually caused the surge. The unemployment rate for young college graduates hit 5.6% in March 2026, up from 3.6% in March 2019, a 56% increase that has generated a steady stream of "AI ate your career" coverage. When researchers tried to isolate the cause, they found something most of that coverage missed: remote work explains 64% of the rise.
AI exposure explained almost none of it.
What Remote Work Actually Does to Entry-Level Hiring
The mechanism matters, because if you understand it you can do something about it.
When a company shifts to distributed work, the calculus for hiring a new graduate changes in a specific way. Training an experienced hire to operate in your environment is a finite project: they come in with skills, you orient them, they're productive within weeks. Training someone who has never worked professionally is a longer and more hands-on process. Managers need to give more frequent feedback, spot mistakes early, answer questions in real time, and model professional behavior. That work is significantly harder when you can't see each other.
"Employers may not want to hire fresh graduates onto distributed teams because it is more difficult to teach them the requisite skills from afar," the NY Fed researchers write. The paper found that feedback on coding work increased 18.3% when workers were physically in the office. Younger workers disproportionately benefited from in-person sessions.
The rational employer response is not to eliminate entry-level hiring entirely. It is to shift the talent you hire toward people who need less development. You post the same job titles but write the requirements differently. You call it "2-3 years minimum" when you used to call it "entry-level." You hire someone who already knows what they're doing because teaching someone from scratch is too expensive when you can't be in the same room.
That's not an AI story. That's a training-model story.
The Fortune 500 Case Study
The paper includes a natural experiment that makes the point cleanly. An unnamed Fortune 500 company in a "remotable" sector went fully distributed after the pandemic and, in the process, shifted its hiring dramatically toward experienced workers: people who could operate independently without mentoring or hand-holding.
Then the company implemented an aggressive return-to-office policy.
New graduate hiring resumed.
The company didn't change its AI investments. It didn't change its business model. The only variable was whether managers could see the people they were trying to develop. When they could, hiring new graduates made economic sense again.
The Awkward Corollary for RTO Debates
This puts recruiters in a mildly uncomfortable position.
For most of the last two years, the evidence on RTO has pointed in one direction: office mandates shrink your candidate pool, increase time-to-fill, and push experienced workers toward remote-first competitors. That is still largely true for senior and specialized roles. A staff-level engineer in a competitive market has options, and requiring five days in the office dramatically narrows who will consider you. Requiring in-office attendance for a director-level finance hire in a city with three competing firms offering hybrid flexibility is a real cost.
But the NY Fed data suggests that for entry-level specifically, the calculus runs differently. Remote work is the structural barrier. Not because companies don't want new graduates, but because the training architecture that makes hiring them worthwhile broke down when everyone went distributed.
Companies that want to rebuild entry-level pipelines can do it. They may just need to reconsider whether fully distributed onboarding is actually compatible with the development investment required to make that hiring pay off.
What the Numbers Say About the Scale of the Problem
The underemployment rate for recent college graduates sits at 41.5%, according to Economic Policy Institute data from Q1 2026. That's the share working jobs that don't require a bachelor's degree. A significant number of recent graduates are not failing to find any job. They are finding jobs that don't match what they trained for, which means their skills are deteriorating in ways that compound over time.
Entry-level unemployment at 5.6% is above the pandemic peak. The NY Fed explicitly notes that early-career experiences have lasting consequences, what economists call "scarring effects." People who start their careers in bad markets earn less for years, change careers more often, and build professional networks that are thinner than their peers who entered in stronger conditions.
This is a problem companies are participating in even when they're not trying to. The decision not to hire a new graduate this year because remote mentoring is hard isn't malicious. It's rational given the constraints. But the aggregate effect is a generation of workers whose human capital is being wasted before it develops.
What Recruiters Should Do With This
Identify clients who recently implemented RTO. The NY Fed finding means companies that have shifted back to in-person or hybrid requirements are more likely to be ready to restart early-career programs. If a client went 5-days-in-office in Q1 2026, that's a tailored conversation about whether their entry-level pipeline needs rebuilding and whether they've updated their job requirements to reflect that.
Stop conflating entry-level and experienced-hire markets. The candidate pool dynamics are different. RTO hurts your candidate pool for senior roles. It may be an enabling condition for entry-level ones. Running the same analysis across both misses the distinction that matters.
Advise on structured onboarding, not just headcount. The reason remote work broke entry-level hiring is that informal mentorship stopped functioning. Companies rebuilding entry-level pipelines need more than headcount: they need a training architecture. A recruiting engagement that includes a conversation about onboarding structure is a more complete solution than one that only delivers resumes.
Watch software development specifically. Employment among developers aged 22-25 has fallen nearly 20% from its late-2022 peak, according to Yale School of Management analysis. Software dev job postings are down 53% from the same point. That cohort is also in the most AI-exposed sector. For tech-focused recruiters, the combination of remote work training barriers and genuine AI displacement of entry-level coding tasks means the headwinds are real and overlapping, not one problem but two.
The Actual Takeaway
The narrative that AI is eating entry-level jobs is not false, but it's incomplete in a way that leads to the wrong interventions. Remote work is the bigger variable for most sectors, and it's one that companies can change without waiting for the AI cycle to play out.
Companies that want to hire and develop early-career talent can do it. The structural fix is not complicated: create conditions where mentoring is possible. What's stopped most of them is an assumption that the problem is external - the AI environment, the talent market, the education system - when a significant portion of it is an internal policy choice they made in 2020 and never revisited.
If you're building or rebuilding an early-career recruiting function, BlueLine's matching tools are designed for the current market. Start at /register.