8 min. read

AI isn’t ending work, but it isn’t leaving work alone either. The best evidence so far shows two things at once: early-career workers in the most AI-exposed jobs are falling behind, and most of the work people do has always been new work. Which way AI cuts inside a company, we’d argue, is decided when someone designs how it enters the workflow.

Key facts

~6 in 10 Jobs people do today that didn’t exist in 1940, by MIT economist David Autor’s estimate. Source: MIT News.

170M / 92M Roles employers expect to be created and displaced by 2030, a forecast from the World Economic Forum’s survey of 1,000+ companies. Source: WEF.

~19% How far employment for 22- to 25-year-olds in the most AI-exposed jobs sits below where it would otherwise be (August 2026). Source: Stanford Digital Economy Lab.

89% Share of companies in BCG’s 2026 survey that expect AI to generate new work; 11% expect it mainly to replace existing jobs. Source: BCG.

The fear, stated fairly

The worry goes like this. AI can now draft, summarize, code, analyze and answer customer questions. Those tasks fill a lot of entry-level jobs. If software does them, companies hire fewer people at the bottom, and the career ladder loses its first rungs. That isn’t a fringe worry. There’s data behind it, and this post takes it seriously.

Most work has always been new work

Start with history. MIT economist David Autor and his co-authors traced U.S. census records back to 1940 and found that most of today’s jobs sit in occupations that didn’t exist then. Autor estimates that “about six out of 10 jobs people are doing at present didn’t exist in 1940” (MIT News).

New work doesn’t arrive evenly, though. Over the last four decades, the research shows clerical jobs shrinking under automation, while new work has landed increasingly in professional jobs for people with degrees and in service jobs for people without them. So “new jobs will appear” is true, and it’s little comfort to someone whose job is the one disappearing.

Many of the jobs that will matter a decade from now don’t have names yet.

Your entry could not be saved. Please try again.

We sent links to your email! You should have received an email from us already. If you did not receive, make sure you check your spam folders and add masterofproject.com to safe senders list to receive our emails.

The forecast: more roles created than displaced

For its Future of Jobs Report 2025, the World Economic Forum surveyed more than 1,000 employers across 55 economies. They expect 170 million roles to be created and 92 million displaced by 2030, a net gain of 78 million. WEF puts the total disruption at 22% of jobs by 2030.

Two cautions. It’s a forecast built from what employers expect, not a count of what has happened. And a net gain doesn’t help the people inside the 92 million.

The hard data: young workers in exposed jobs are behind

One careful look at what has actually happened comes from Stanford’s Digital Economy Lab, where Erik Brynjolfsson, Bharat Chandar and Ruyu Chen track payroll data by age and by how exposed each occupation is to AI. Their August 2026 update finds:

Young workers are behind. Among 22- to 25-year-olds in the occupations most exposed to AI, employment sits about 19% lower than it would if it had matched the trend for same-age workers in less-exposed fields. A year earlier, the gap was 15%.
Experienced workers aren’t. Older workers in the same occupations show no comparable gap.
How AI is used matters. The drops are concentrated in occupations where AI mostly takes over tasks. Where AI mostly assists people, employment has held steady or grown, especially for experienced workers.

The authors are careful about what this means. They see no widespread, economy-wide job displacement associated with AI. They describe these as patterns, not proof that AI caused them. The gaps shrink once education is taken into account, and some of the trends began before generative AI was widely used. Still, the gap for young workers is real, and it has widened over the past year.

Automate or augment is a design choice

Here’s how we read the Stanford finding. The research measures whole occupations, not individual companies, so this is our interpretation rather than the authors’ conclusion. The same technology shows up in both kinds of occupation. What differs is how it’s used: to take over the task, or to help the person doing it. Inside a company, that choice gets made during deployment, workflow by workflow, by whoever designs how AI enters the work.

Most companies don’t make the choice deliberately. A tool gets licensed, a vendor builds what was scoped, and the effect on jobs is whatever falls out. An AI-native company makes it on purpose: it decides which tasks to automate, which to augment and which to keep human, and it names someone to own the result.

BCG’s Applied AI Index 2026 adds a related signal from the employer side. Of the companies it surveyed, 89% expect AI to generate new work, against 11% who expect it mainly to replace existing jobs. Expectations aren’t outcomes, but they show where leaders think the work is going.

A job that barely existed five years ago

Here’s one example of the new class of work. A few years ago, few organizations had a job whose purpose was owning how AI gets deployed: deciding what a system may do on its own, where people stay in the loop, how its results are measured and what happens when it drifts. Today that job is the AI project manager, and in September 2026 Maryland registered an apprenticeship for it under the Information Technology Project Manager occupation.

Master of Project Academy sponsors the nation’s first registered apprenticeship dedicated to AI Project Management, registered by the Maryland Apprenticeship and Training Council (MATC #2575, September 8, 2026) under the registered occupation Information Technology Project Manager (O*NET 15-1299.09). The registration was announced via Business Wire.

It’s one example. BCG found that the most AI-mature companies in its survey already have 13% of their employees in dedicated AI roles. Many of those roles are still settling on their names.

If AI thins the bottom rungs, build them on purpose

The Stanford data says the pressure lands hardest at the start of careers. That’s the part companies can do something about. If AI takes over the tasks that used to train new hires, the training has to be rebuilt deliberately: paid roles where early-career people learn by doing real work alongside someone experienced.

That’s the old logic of apprenticeship: earn while you learn, on the job, with a mentor. It fits the new AI roles especially well, because owning how AI enters a real workflow is hardest to learn anywhere but on real workflows. The AI Project Management apprenticeship pairs 4,000 hours of mentored on-the-job learning with 340 hours of related instruction over 24 months. The apprenticeship hub explains how it works, and our step-by-step guide to AI job redesign walks through the redesign itself.

Two questions for any AI rollout

For each workflow AI is entering this quarter: did someone decide, task by task, what to automate and what to augment, or did it just happen?

And if AI now does the work your entry-level people used to learn on, where will next year’s experienced people come from?

Frequently asked questions

Will AI replace project managers?

AI is automating parts of project administration and creating a new kind of project manager: one who owns how AI is deployed inside the business. That’s a role being added, not replaced. In September 2026, Maryland registered an apprenticeship for it under the Information Technology Project Manager occupation.

What new jobs is AI creating?

Roles that design, run and oversee AI inside organizations, such as AI project managers who own deployment, alongside roles in AI operations, evaluation and governance. BCG found the most AI-mature companies in its 2026 survey already have 13% of their employees in dedicated AI roles. Many of these jobs don’t have settled titles yet.

Which jobs are most affected by AI so far?

Stanford’s Digital Economy Lab finds the clearest effect among workers aged 22 to 25 in the most AI-exposed occupations, where employment is about 19% below where it would otherwise be (August 2026). Experienced workers show no comparable gap, and declines concentrate where AI automates tasks rather than assisting people. The authors describe these as patterns, not proof of cause.

Build the entry rung on purpose

The AI Project Management apprenticeship, registered under the Information Technology Project Manager occupation (O*NET 15-1299.09; MATC #2575), lets you grow an AI-native project leader from early-career talent or your own team, paid and on the job from day one.

Book a 30-minute curriculum review →

P.S. Looking for an AI project manager role yourself? Our guide to how to become an AI project manager covers both ways in.

About Master of Project Academy

Founded in 2012, Master of Project Academy has served 500,000+ enrollments across 180+ countries in project management training. It sponsors the AI Project Management apprenticeship, registered under the Information Technology Project Manager occupation (MATC #2575). Learn more at masterofproject.com/p/aipmapprenticeship.

Sources

AI Isn’t Ending Work. It’s Creating Jobs That Don’t Have Names Yet