Here’s a stat that should stop you mid-scroll. Since 2021, advertised pay for the most AI-exposed jobs in the US has climbed 46%. Meanwhile, artificial intelligence remains the single most-cited reason employers give for cutting jobs this year. So the same technology, in the same labor market, is producing two very different outcomes. That depends on where you sit in it. If you run a business, or manage any team that touches AI, this split is not a curiosity. Instead, it is your next hiring plan, whether you have written it down yet or not.
What actually happened this week
On September 17, Indeed’s Hiring Lab published new research built on the company’s own job posting data. The finding is straightforward. For example, ChatGPT and Claude are generative AI tools that can write, summarize, and analyze on command. In jobs where a tool like that touches a meaningful share of daily tasks, pay has grown 46% since 2021. That still compares with just 25% for the least AI-exposed jobs, things like nursing, food prep, and manufacturing.
Meanwhile, outplacement firm Challenger, Gray and Christmas released its August 2026 job cuts report. Through August, AI has been cited in 116,175 announced layoffs this year, about 22% of every job cut nationwide. In addition, it was the single leading reason for job cuts for five straight months, from March through July.
The number under the number
Dig one layer deeper and the story gets sharper. Indeed also ran the numbers a different way. Once you control for which occupations and seniority levels are posting jobs, that changes things. The raw AI pay premium turns out to be just 5.7 percentage points. Control further for seniority specifically, and it narrows to 2.4%. So part of that headline 46% is not really about AI paying more for the same work. Instead, it is largely about which jobs are even being posted anymore.
Here’s the part that matters most. The most AI-exposed fields include software development, IT systems, data and analytics, marketing, and banking and finance. In those fields, entry-level postings fell from 29% of all listings in 2021 to just 10% today. Senior-level postings, meanwhile, rose from 22% to 47% over the same stretch.
Read that again. Your industry is not simply paying AI-savvy people more. At the same time, it is quietly closing the door. That door used to let junior people become AI-savvy people in the first place.
What most of the coverage is getting wrong
Most of what you have read on this treats the pay story and the layoff story as two separate headlines. One says AI is good for your paycheck. The other says AI is coming for your job. Both framings miss the actual mechanism.
This is not two trends running side by side. It is one trend instead. That trend is a bifurcation, meaning the workforce is splitting into two groups instead of moving up together as one. A shrinking group of experienced, AI-fluent specialists is getting paid more because they are now scarcer and more valuable. Meanwhile, a shrinking on-ramp is making sure fewer new people ever reach that group.
Also worth noting: in August, restructuring, not AI, led the list of cited reasons for the first time since February. So check your own language first. Before you label every layoff in your sector an “AI layoff,” see what your own leadership is calling it internally. Words matter here, especially to your own people.
Why this is your problem, not just an economist’s chart
You do not need to be a Fortune 500 company for this to apply to you. First, think about how you have hired in the past. Maybe you brought on a junior analyst, a first-year developer, or an entry-level marketer. If you had them learn on the job through small, repetitive tasks, you were running a training pipeline. It just didn’t have that name. AI can now do a lot of those small tasks. So the real question is not whether AI can do them. It is whether you assign them to a human anyway, because that is how your future senior people get made.
Skip that step for a year or two and it looks free. Skip it for five years, however, and the bill comes due. You will end up paying a 46% premium to hire senior talent. That’s talent you used to grow yourselves, before nobody in your pipeline got the chance to become senior.
What to do about it this quarter
This is exactly the kind of decision my A.S.K. Framework was built for. It sorts every task into three buckets: Automate, Share, or Keep Human. That way, AI augments your people’s thinking instead of quietly replacing the tasks that build it.
Here is where I would start. Right?
- Audit your entry-level roles for what is still left in them. If AI has absorbed every task that used to teach a junior hire the job, that role isn’t entry-level anymore. It is a senior role wearing an entry-level salary, and that is not a stable place to hire from.
- Protect at least one real “learning task” per junior position, on purpose. Not busywork. Pick a task that moves slower with a human doing it, one that builds the judgment your manager will need. Keep it human, even if a tool is faster.
- Separate “AI” from “restructuring” in your own layoff language. Challenger’s own August numbers show restructuring, not AI, led the list of reasons for the first time since February. So if you are cutting people, name the real reason. After all, your remaining team is watching how honestly you talk about this.
- Watch your own pay data the way Indeed just watched the whole country’s. If your AI-fluent specialists are getting raises while your junior pipeline shrinks, that is not a coincidence. Instead, it is a strategy, whether anyone chose it on purpose or not.
- Budget for mentorship the same way you budget for tools. A senior hire who can prompt, verify, and correct an AI system did not learn that from a course. They learned it from years of doing smaller versions of the job under someone who checked their work.
The part I want you to sit with
None of this means slow down on AI. The 46% premium tells you the market already rewards people who use it well. Instead, it means being deliberate about who gets the chance to become one of those people next. The organizations that are fine in five years will not be the ones that automated fastest. They will be the ones that kept training a bench while everyone else was busy shrinking theirs.
Want to talk through what your own entry-level pipeline looks like right now? Hit reply, or drop a comment below. I read every one of them.
Así que sí, mi gente, la inteligencia artificial le paga bien a quien ya sabe. La pregunta de verdad es quien le va a enseñar al que todavia no sabe. ¡Wepa, a seguir formando gente!
Sources
- Indeed Hiring Lab, “AI Exposure Isn’t Squeezing Advertised Pay in the US, It’s Boosting It” (September 17, 2026)
- Challenger, Gray and Christmas, “Challenger Report: August Job Cuts Up 58%, Consumer Products, Food Lead” (September 2, 2026)
- 24/7 Wall St., “AI Is Hollowing Out Software Jobs, But Paying Survivors a Fortune” (September 22, 2026)
- Outsource Accelerator, “AI-exposed jobs see faster pay growth: Indeed” (September 23, 2026)
Cover photo by Luca Bravo on Unsplash. In-post photo by Brusk Dede on Unsplash.