The Best Internship on That Page Has a Weird Title

In partnership with

Trying to figure out a career is hard enough without all the noise around you. Shortform helps it make sense with in-depth guides that go beyond plain summaries. It breaks books down chapter by chapter and connects ideas across authors who’ve tackled the same questions.

Their guide How to Start by Jodi Kantor is a great place to begin, pulling in perspectives from Designing Your Life and The Squiggly Career along the way.

It’s part of a library spanning multiple genres including Career, Self-Improvement, and Productivity. New guides are added weekly and subscribers vote on what’s next.

Try it free and save 49% on the annual plan

Welcome to today's SCALIS EarlyCareers newsletter! 🚀

It is September, which means you are on a careers page right now with the word intern typed into the filter box. You scan the results, recognize four titles, and apply to those four. Software Engineer Intern. Marketing Intern. Financial Analyst Intern. Summer Analyst.

Three rows down sits a posting called something like Applied AI Intern, Evaluation, or AI Solutions Intern. You scrolled past it. Not because you read it and decided against it, but because it did not look like a job you were allowed to have.

Here is what the data says about that posting. According to NACE's Job Outlook 2026 Spring Update, more than a third of entry level jobs now require AI skills, nearly triple the share employers reported six months earlier. Twenty eight percent say they are specifically seeking early career talent who can use AI. Nearly sixty percent say they are assigning interns projects that use AI tools.

A category cannot grow that fast and also have a settled applicant pool. There is no feeder school for a job title that is eighteen months old, no returning intern class, no upperclassman who did it last summer and told everyone how. The gate that keeps you out of the crowded roles has not been built here yet.

Make Your Screen Time More Productive

Instead of wasting hours on mindless content, spend 10 minutes learning something that can actually improve your life.

Shortform distills the key ideas from the world’s best nonfiction books, articles, and podcasts into concise, actionable summaries.

Turn 10 minutes a day into a smarter, more capable you.

Get 49% OFF

3,874 BRAND NEW internship positions freshly posted today. View the full list below and start applying!

The experience line on a new role is arithmetic fiction

Open one of those postings and you will find a requirement like two or three years of experience with agent frameworks, model evaluation, or retrieval systems. Read that as arithmetic. Those tools have existed at production scale for roughly that long in total. The number of undergraduates on earth who clear that line is approximately zero, and the recruiter who posted it knows.

What happened is mundane. Someone copied the requirements block from an adjacent senior posting, because the role is new and there was no template for it. On an established internship, requirements are a filter. On a role invented last year, they are a wish list written by a manager still figuring out the job.

The students who get these seats are not more qualified than you. They understood the category is too young to have qualified people in it, and applied anyway.

Search the work, not the title

The reason you are missing these postings is the filter box. You are searching titles, and the titles are unstable. The same job gets posted as Applied AI, GenAI, LLM Engineering, Forward Deployed, AI Solutions, and Evaluation, sometimes at one company in one quarter, because nobody has agreed on the vocabulary yet.

So search verbs instead. On any careers site, run the keyword search on the work: evaluation, prompt, fine tune, agent, annotation, retrieval, benchmark. Then drop the word intern from the query entirely and sort the results by the eligibility line instead. A surprising number of these reqs never got tagged as internships, because the person who created them was not a campus recruiter and did not know the tagging convention.

Do this on five target employers tonight. You will surface postings your classmates, all typing the same four title words into the same box, will never see.

Most of these seats are not at AI companies

This is the part students get backwards. You hear AI internship, picture a frontier lab, decide the odds are hopeless, and stop looking.

Go back to the NACE number. Nearly sixty percent of employers say they are assigning interns projects that use AI tools. That is not sixty percent of AI labs. That is insurers, hospital systems, retailers, regional banks, manufacturers, school districts. The AI work is already sitting inside internships with entirely ordinary titles.

Which means the posting you want may be called Marketing Intern, with the AI in the fourth bullet of the responsibilities section. Read the bullets, not the header. A conventional internship with an AI paragraph buried in it is a seat where your fluency is a differentiator rather than the price of entry, and where the pool is still whoever applies to a marketing internship. If you are not a computer science major, this is your lane, and it is the least contested one on the page.

Make Your Screen Time More Productive

Instead of wasting hours on mindless content, spend 10 minutes learning something that can actually improve your life.

Shortform distills the key ideas from the world’s best nonfiction books, articles, and podcasts into concise, actionable summaries.

Turn 10 minutes a day into a smarter, more capable you.

Get 49% OFF

Learn which AI doors are shut while you are enrolled

This category borrows vocabulary campus recruiting does not use. Residency. Fellowship. Apprenticeship. Studio. These are real paid entry level doors, and several are closed to you specifically because you are a student.

The OpenAI Residency says so on its own page: residents are full time employees and cannot be actively enrolled in any school program. The page then tells students what to do instead, which is to wait and apply roughly three months out from finishing their studies. That is a graduation year door, not a sophomore summer door, and every application you send there while enrolled is one you did not send somewhere live.

Build one more habit while you are at it. Check the date on any program page before you plan a semester around it. Meta's AI Residency page is still live, still indexed, and still says applications are closed for a cohort that started years ago. Program pages rot and nobody takes them down. If a page does not name a current cycle, treat it as a museum.

The window closes one cycle at a time

This advantage is not permanent, and it does not expire all at once. It decays by one recruiting cycle.

Right now a brand new title has no incumbent comparison, because no student did that exact internship last summer. Next fall there will be one. The year after, a returning cohort, a defined pipeline, a short list of schools the team learned to recruit from, and a resume shape that reads as standard for the role. That is how every category hardens. Cloud did it. Mobile did it. Data science did it. Each was briefly wide open and then was not.

You are reading this during the open part.

The twenty minute pass, and the email after it

Pick five employers you would actually work for. On each careers page run three searches: the word intern, then evaluation, then agent. Write down every posting that shows up in the second and third but not the first. That list is your edge.

Then, for the one you want most, send this to the recruiter named on the posting or to the campus recruiting address:

That does two jobs. It gets you real information about a role whose posting is vague because the work is still being defined. And it marks you as the candidate who noticed the role was new, which is the instinct the team is hiring for in the first place.

Learn AI in 5 minutes a day

You don't have to scroll every AI thread, track every new tool, or watch every demo. 

The Rundown AI breaks it all down for you — the latest AI news, tools, and tutorials in one free 5-minute email every morning. 

Trusted by 2M+ professionals at Apple, Google, and NASA.

Sign up to start learning.