Jobs AI Is Already Replacing (2026-2027 Update)
Forget predictions for a second. Here's what's actually documented: real companies, named, cutting real roles and pointing at AI.
Forget predictions for a second. Here's what's actually documented: real companies, named, cutting real roles and pointing at AI. The clearest cases so far are in customer service, back-office support, and routine coding, and the pattern is consistent: the roles going first are the high-volume, routine ones, not the whole occupation. This piece sticks to verified cases and hard numbers, no speculation.
For the bigger picture of which roles are most exposed and why, see the pillar, Which Jobs Will AI Replace First? Here we're only looking at what's already happened.
Customer service: the clearest case so far
Customer service is the most-documented example of AI replacing real jobs, and its story also contains the biggest cautionary tale.
Klarna is the headline case. The fintech cut roughly 700 customer service roles as its AI assistant took over routine support, with the CEO publicly claiming in 2024 that the AI was doing the work of 700 agents (Fast Company). Then the twist: by 2025, with customer satisfaction slipping, Klarna reversed course and began rehiring, moving to a hybrid model where AI handles high-volume routine queries and humans take escalations and complex cases (digitalapplied).
Read that carefully, because it's the whole story of "what jobs is AI replacing" in miniature. AI genuinely absorbed the routine tier of customer service: the repetitive, scriptable tickets. It could not absorb the hard tier: the frustrated customer, the messy edge case, the judgment call. Klarna's reversal wasn't AI failing at customer service. It was AI succeeding at part of customer service and the company overreaching on the rest. The routine slice got replaced and stayed replaced. The human slice came back.
Back-office and HR: quieter, but real
IBM offers the clearest back-office example. CEO Arvind Krishna told the Wall Street Journal that AI had taken over the work of "several hundred" HR employees, and in November 2025 IBM said AI agents had already replaced hundreds of back-office roles, with further cuts of roughly 2,000 to 3,000 as it shifted people toward AI and quantum work (programs.com AI layoffs tracker).
The tasks here fit the pattern exactly: routine HR processing, standard back-office administration: structured, repetitive, digital. Note what IBM did not say it automated: the judgment-heavy HR work, the sensitive personnel decisions, the strategy. The replaced roles were the process-execution tier.
Coding: the entry-level squeeze
Software isn't immune, but the effect is concentrated at the bottom rung rather than across the board.
The Stanford payroll study found entry-level software engineering roles fell by nearly 20% between late 2022 and mid-2025, while experienced engineers held steady or grew (TIME). This isn't companies announcing "we replaced our developers with AI." It's quieter: fewer junior hires, because AI now covers a big share of the boilerplate coding and simple fixes that used to be a junior's first year of work. The job didn't get eliminated by press release. The bottom of the ladder just got shorter.
The numbers behind the trend
Individual cases are vivid, but the aggregate matters more, and it's climbing fast.
Through the first half of 2026, AI was cited in 101,743 US job cuts, nearly double the 54,836 attributed to it across all of 2025, according to Challenger, Gray & Christmas data reported by industry trackers (Founder Reports AI layoffs tracker). That's a steep acceleration in roughly a year.
Two honest caveats on that number. First, "AI-attributed" is squishy: some companies blame AI for cuts that were really about cost-cutting or over-hiring during the pandemic boom, because "restructuring for AI" sounds better to investors than "we hired too many people." Reporting has flagged that AI's stated role can be both over- and under-counted depending on the company's incentives (CNBC). Second, cuts aren't the same as net job loss: the WEF still projects net job growth through 2030, even with heavy churn. So take the acceleration seriously as a signal, without treating every AI-attributed layoff as a clean case of a robot doing someone's exact job.
The through-line: routine tasks, not whole professions
Put the documented cases side by side and one pattern holds across all of them:
| Where AI replaced work | The tasks it took | The tasks it didn't |
|---|---|---|
| Klarna customer service | Routine, scriptable tickets | Escalations, complex and emotional cases |
| IBM back-office / HR | Standard processing, admin | Sensitive decisions, judgment, strategy |
| Entry-level coding | Boilerplate, simple fixes | Architecture, senior problem-solving |
Nobody in these documented cases replaced an entire profession. They replaced the routine, high-volume, learnable tier, and in Klarna's case, overreached and had to walk part of it back. That's the practical meaning of "what jobs is AI replacing" in 2026: the automatable slice of exposed roles, first and fastest, with the human-judgment slice proving stubbornly hard to hand over.
Which is exactly why your title doesn't tell you your risk. Your task mix does. If your role is mostly the routine tier that got replaced at Klarna and IBM, that's a real signal to reposition. If it's mostly the tier that came back, you're steadier than the headlines suggest.
What this means for you
Don't read a layoff headline and assume your job is next because you share a title with someone who got cut. Read it and ask: how much of my week is the routine tier that actually got automated in these cases? If it's a lot, start shifting your time toward the judgment, relationship, and accountability work that kept its value in every one of these examples, and get fluent with the AI tools in your field so you're operating them, not competing with them.
The fastest way to see where you stand is to score your own tasks. The free, anonymous How AI-Proof Is Your Job? assessment does exactly that. And if you want to know how people describe what it's actually like to have a slice of their job automated, our reader piece AI Took Over Part Of My Job is worth a read.
Frequently asked questions
What jobs is AI replacing right now?
Documented cases center on routine customer service (Klarna cut ~700 support roles), back-office and HR processing (IBM automated the work of several hundred HR staff), and entry-level coding, where junior software roles fell nearly 20% from late 2022 to mid-2025 (Fast Company; TIME).
How many jobs has AI actually cut?
AI was cited in 101,743 US job cuts in the first half of 2026, up from 54,836 in all of 2025 (Founder Reports). But "AI-attributed" figures are imperfect: some companies overstate or understate AI's role (CNBC).
Did any company reverse its AI layoffs?
Yes. Klarna, after cutting ~700 customer service roles for AI, began rehiring as satisfaction dropped and moved to a hybrid AI-plus-human model (digitalapplied). AI held the routine tier; humans came back for the complex work.
Is AI replacing whole jobs or just parts of them?
In every well-documented case so far, AI replaced the routine tier of a role, not the whole profession. The judgment, relationship, and accountability parts have proven much harder to automate.
How do I know if my job is one AI is replacing?
Look at your tasks, not your title. If most of your week is routine and digital with low accountability, you're in the tier that's being automated. The free How AI-Proof Is Your Job? assessment scores this for you.
See where your own role lands. Take the free, anonymous How AI-Proof Is Your Job? assessment in about five minutes.
Published August 2026.