The Same AI That Could Replace Your Job Might Also Give You a Raise
Sign Up For Daily Job Market News
A new Morgan Stanley note is making the rounds this week, and it argues something that cuts against most of what you’ve read about AI and jobs. Economist Heather Berger’s team says the workers most exposed to AI displacement are often the same workers positioned to gain the most from it. Berger’s team calls them CHIC: college-educated, high-income, city-dwelling households.
Berger names four channels working in their favor: productivity-driven wage growth, new job creation, wealth effects, and longer-term disinflation. In her team’s words, “We think these upside channels are underappreciated.” The idea is simple even if the acronym isn’t: white-collar workers using AI well tend to get paid more for the output, and everyone in that income bracket ends up living in a slightly cheaper economy as a result.
The age split is where this gets more specific. Younger CHIC workers face a real risk of losing the routine, entry-level tasks AI already handles well. Older colleagues in the same fields are more likely to see wage growth from the same technology without losing their jobs to it. Berger’s team is candid that job creation hasn’t caught up yet. “In terms of job creation, it is still early, but new AI-related occupations have so far been aimed at these same CHIC consumers,” with new postings “targeted towards higher-income consumers with some experience in similar ‘highly-exposed’ industries.”
The piece also brings in Andrew Slimmon, head of applied equity advisors at Morgan Stanley Investment Management, who earlier this year said he wasn’t worried about AI’s employment impact, comparing the expected rebound to how the early-2000s dot-com boom reshaped the labor force instead of shrinking it.
Here’s what I’d add from where I sit, reviewing requisitions every day. The age split matches exactly what I see: clients are paying up for senior people and cutting way back on junior hires. And Berger’s own data undercuts the optimism more than her framing lets on. If new AI jobs are going to people who already have experience in AI-exposed industries, that’s a story about the same experienced people getting paid twice, once for their old skills and again for adding AI to them, not about displaced workers landing on their feet. An investment bank telling high-income households they’ll come out ahead is also a note about consumer spending, since high earners are the ones who drive it. The practical read for anyone job hunting right now: wage growth is going to people who can point to what they built with AI tools, and it’s passing by people who quietly hoped it wouldn’t touch their desk.
Staffing your team doesn’t have to be hard.
Reach out and see how we can help.
Work Trials Are Replacing the Resume as the Real Screening Test
AI didn’t just change which jobs are exposed to automation. It also broke the resume as a screening tool, and employers are responding by asking candidates to prove themselves before they get an offer. Morning Brew reports that companies are leaning harder on “work trials,” direct auditions where a candidate does a version of the job itself, because AI-optimized resumes and cover letters have made real applications hard to tell apart from generated ones.
The format ranges widely. On one end is an unpaid, off-site mock task that mirrors the job. On the other is a full week in the office working alongside the team the candidate would join. The rule that should govern all of it is simple and, per Morning Brew’s reporting, often ignored: if the work benefits the company, the candidate generally has to be paid for it.
The scale of this shift is bigger than the headline trend piece suggests. A 2025 survey from the National Association of Colleges and Employers found that roughly two-thirds of companies already use some form of skill-based hiring for entry-level roles. Work trials are the most visible version of a change that’s already mainstream.
I like the instinct behind this and think a lot of employers are getting the execution wrong. AI wrecked the resume as a reliable filter, so employers need real proof of ability; that reasoning is sound. But plenty of companies are reaching for the most expensive version of that proof, a multi-day or week-long trial, when a 90-minute exercise would tell them almost everything they need to know. The math doesn’t even work in the employer’s favor. A week-long trial filters out currently employed candidates who can’t take unpaid time away from their current job, leaving mostly people between jobs or people willing to burn PTO on a maybe. Neither group is a reliable proxy for who’s the best hire, and I’ve watched clients lose strong, employed candidates to exactly this problem. There’s also a wage-and-hour legal issue buried in the unpaid version of this. If the work has value to the company, treating it as a favor to the candidate is a real liability.
My advice to hiring managers: cap the trial at an afternoon, pay for it, share your scoring rubric up front, and give every candidate feedback whether or not they get the offer. My advice to candidates: ask what you’re being evaluated on and whether you’ll be paid before you agree to anything longer than a couple of hours. If your current hiring process still leans entirely on resumes and gut-feel interviews, or you want help designing something more rigorous without asking candidates to give up a week of their life, talk to us about hiring and we’ll help you build a process that predicts performance.
A Simple Test for Whether AI Is Coming for Your Job
London Business School professor Lynda Gratton, in an excerpt from her new book “Living the 100-Year Life,” offers a way to think about your own exposure that’s more useful than most of what’s floating around online. List roughly 30 things you do at work in a typical week, then sort each one along two lines: routine or nonroutine, and manual or cognitive. The result is four boxes, and each one has a different relationship with automation.
Routine, manual work is the most exposed category. Gratton’s own first job in 1970, packing chocolates on a conveyor belt, was fully automated within a decade. Routine, cognitive work follows the same pattern on a lag. She describes doing statistical analysis on a university mainframe in 1980, work that later got automated the way spreadsheet assistants and bank tellers eventually did.
Nonroutine manual work, like driving or plumbing, was supposed to be next. Gratton’s 2010 book “The Shift” predicted driverless cars and robotic plumbers by 2025. The timeline missed, mostly because of safety regulation and cost, not the technology itself, though Waymo now operates driverless taxis in San Francisco and money is pouring into robotics startups like Figure AI.
The box that flipped first, starting in 2023, was nonroutine cognitive work, the category everyone assumed was safest. Generative AI now writes essays, provides one of the most common uses of AI tools today in mental health support, and coaches people through performance-review prep. Gratton makes the point personal: ChatGPT could reportedly write “an essay like Lynda Gratton” because it had ingested her books, articles, and blogs without her permission.
Gratton’s advice is to build resilience and make reflection a habit, since the cycle of reassessing your own value has gone from once a decade to something closer to monthly. I’d push that further with a version of her own exercise: don’t just sort your 30 tasks into boxes, put a rough number of hours next to each one. Most jobs hold something like 70% routine cognitive work wrapped around a handful of hours of real judgment calls, and that 70% is exactly what’s leaving first. Her admission about missing the 2010 prediction is the most honest part of the piece. Forecasters expected blue-collar work to get hit first. Instead the software went after essay writers, analysts, and coaches, which tracks with what I’m hearing from hiring managers right now: companies that planned to trim warehouse headcount are instead cutting marketing coordinators and junior analysts. The practical version of Gratton’s advice is moving your own hours out of box two and into box four: deciding, persuading, taking responsibility – the tasks companies still pay a human to do.
