AI Is Squeezing Entry-Level Hiring. Here's How Juniors Get In

If you're a graduate, a career-changer or a junior engineer trying to land your first AI or automation job right now, you're not imagining it: the bottom rung of the ladder has got harder to reach. A new survey from Northeastern University puts real numbers on what a lot of young candidates have been feeling, and the picture is mixed. Fewer entry-level hires at some companies, higher expectations at others, and software reading your résumé before a human ever sees it.

Here's the thing, though. I've read the data closely, and I don't think the message is "don't bother". I think the message is "show up differently". The juniors who treat this as a new game with new rules are going to do very well, because the same AI that's squeezing the old entry-level jobs is creating a whole new kind of junior role. Let me walk through what the research actually says, then get practical.

What the Northeastern survey found

Northeastern's AI & Society Survey, reported on 2 November 2026, asked 8,174 people who have a say in hiring how AI is changing their decisions. Roughly half said AI had made a difference. Among that group, 26% said they were hiring fewer entry-level candidates, while only 7% said they were hiring more. The rest kept their hiring steady, but 28% said they now expect more output or higher-level work from entry-level employees.

The screening side is just as striking. AI is on résumé duty for 54% of hiring decision-makers, rising to 83% in tech. And the volume problem is real: the report notes that the number of applications companies receive has more than doubled since 2022. As one employment lawyer quoted in the piece put it, this is happening on both sides of the table. Candidates use AI to fire off more applications, so employers use AI to sort the flood.

The line from that article I keep coming back to is from Gary Friedman, an employment trial lawyer at Seyfarth Shaw: "AI can hollow out the bottom rung of the ladder or it can turn it into an escalator." That's exactly the choice in front of every junior right now.

There's good news buried in there too. Around 40% of hiring decision-makers saw AI use on every writing or coding assignment as a plus, against only 21% who said it would put them off making an offer. And almost half said AI experience wasn't explicitly on their radar, which means candidates who can show it still stand out in plenty of industries.

The bigger picture: the entry level is being "seniorised"

Northeastern isn't the only one seeing this. PwC's 2026 Global AI Jobs Barometer, which analysed more than a billion job ads, looked at 2.4 million US entry-level jobs and found that the roles most exposed to AI are now seven times more likely to require traditionally senior skills like judgement, leadership and face-to-face work. The kicker: those "seniorised" entry-level roles have grown 35% since 2019, while other entry-level roles shrank by 10%, according to PwC's launch release.

Read that again, because it's the most important number in this post. The entry level isn't disappearing. It's changing shape. Junior jobs that ask for more are growing. Junior jobs that only ask you to do routine work are the ones shrinking. PwC's Pete Brown summed it up well: AI is removing some of the routine work that once acted as an apprenticeship, while increasing demand for judgement, leadership and adaptability much earlier in careers.

The same report found jobs requiring specific AI skills growing about eight times as fast as the overall jobs market (69% versus 9%), with an average wage premium of 62% for AI skills. So the prize for getting in is bigger than ever.

In the specialist world of agentic AI, the squeeze is sharper. AgenticCareers' State of Agentic Hiring for November 2026 counts 2,949 open agentic roles, but only 114 of them are junior or internship level (56 junior, 58 internships), against 454 staff-level listings. Their explanation is one I agree with: agent teams are small and new, and they're mostly being built by moving experienced engineers across, which leaves nobody whose job is to grow a junior. They call it a temporary condition and point to internships as the leading indicator. They also found something useful for job hunters: the entry-level roles that do exist cluster with the smallest employers, not the biggest.

How juniors break in now

So what do you actually do? Here's the playbook I'd follow if I were starting out in AI and automation today.

1. Build a portfolio of shipped agents, not certificates. If employers expect higher-level work from day one, prove you can already do it. Build two or three small AI agents that do a real job end to end: one that triages a support inbox and drafts replies for approval, one that reconciles a spreadsheet against an API, one that pulls data from a public source and writes a weekly summary. Put them on GitHub with a clear README, a short demo video and an honest "what broke" section. Tools like Grok Bot, Claude or the OpenAI platform are all fine places to start; pick the one you're comfortable with and focus on the system you build around it.

2. Make evaluation your signature skill. This is the single fastest way to look senior as a junior. For each project, write a test set of 30 to 50 real tasks, measure success rate, cost and latency, then show how you improved them. Most applicants say "I built a chatbot". You'll be the one saying "my agent went from 72% to 94% task success and I halved the cost per run". That's the "higher-level work" those 28% of hiring managers are looking for.

3. Contribute to open source. Agent frameworks, evaluation libraries and MCP tool servers are all open and actively maintained. A handful of merged pull requests, even documentation fixes and small bug fixes, gives a hiring manager public proof that you can work in a real codebase with real reviewers. It also gets you known by the exact people who end up hiring.

4. Look hard at apprenticeships, graduate programmes and internships. Structured programmes are where companies still deliberately invest in juniors. On our own board right now there are roles like Lenovo's Junior Data and AI-ML Engineer, Magna's Junior AI Engineer, a Junior AI Developer role for WWF, Adobe's 2027 University Graduate Machine Learning Engineer and an American Express 2027 AI Engineer internship. Apply early. These programmes often fill months ahead.

5. Use adjacent roles as a side door. Not every route goes straight through "AI engineer". AI operations, data analyst, automation specialist, technical support for an AI product, QA and evaluation, implementation consultant and solutions engineering are all real ways in, and several of them reward domain knowledge. If you've worked in payroll, logistics, claims or customer service, you understand the job an agent is meant to do better than most engineers. That's an asset, not a gap.

6. Go small and go early. Given the AgenticCareers finding that entry-level agentic roles cluster at small employers, don't just queue up behind thousands of applicants at the biggest names. A ten-person startup with one AI role can be a much faster door in, and you'll learn more in a year there than in three years doing narrow work somewhere huge.

7. Write for the robot, then win the human. With AI screening most tech résumés, use clear role titles, plain section headings and the actual skills listed in the ad, written honestly. Then put your best evidence (links to shipped projects and eval results) right at the top so the human who does read it is impressed in ten seconds. And don't spray hundreds of generic applications. That's exactly the flood screening tools were built to filter. Fewer, sharper, tailored applications win.

8. Lean into the human skills. PwC's data says the growing junior jobs demand judgement, leadership and communication. So show them. Write clear project write-ups, present a demo at a local meetup, mentor someone a step behind you, run a small team project. Northeastern also found around 38% of hiring decision-makers say a degree makes a candidate more valuable, and the survey's co-author links that to the critical thinking a degree builds, so if you have one, make that thinking visible.

A note for employers

If you're hiring, I'd gently push back on cutting juniors. Every senior engineer you'll want in 2030 is a junior today, and if the whole industry stops training them we'll all be fighting over the same small pool of experienced people, at ever higher salaries. PwC's research suggests the winners are redesigning work around AI, not just automating tasks away, and that's exactly where a sharp, AI-native junior shines.

Practically, a few things help. Rewrite entry-level roles around real tasks rather than years of experience. Swap some résumé screening for a short, paid work sample. Pair juniors with a senior and give them ownership of something small but real, like an internal agent or an evaluation suite. And check your screening tools regularly, because as Friedman warned in the Northeastern piece, an algorithm can make thousands of mistakes in a nanosecond. You might be filtering out exactly the person you need.

My take

I'm optimistic. Yes, the old entry-level model of "learn the routine stuff for two years, then get trusted with real work" is fading. But it's being replaced by something that, for motivated juniors, is arguably better: the chance to do meaningful, high-leverage work from week one, with AI as a co-worker. The escalator is there. You just have to step on it with proof in hand.

Ready to start? Browse live junior, graduate and early-career roles in AI and automation on our jobs board, pick one that stretches you, and build the project that shows you can do it.