30/05/2025
Two years ago a friend of mine — let’s call her Ana — worked in the back‑office team of a midsize Brazilian bank. Her desk was buried in onboarding forms, ID scans, and the daily tedium of copy‑pasting numbers into spreadsheets. Then her manager rolled out an AI tool that could read documents, fill the right cells, and verify signatures in seconds. Within six months half the team had quietly left, their seats unfilled. “Nobody actually got fired,” Ana told me. “They just… stopped replacing people.”
That story isn’t unique. Since ChatGPT’s splashy debut in late 2022, bosses everywhere have been asking the same question: Can fewer people do more if we bolt AI onto our workday? The data says yes—up to a point. But there’s nuance (and the occasional fiasco).
A headline about thousands of layoffs grabs attention, but the quieter story is hiring freezes. IBM’s CEO famously hit pause on 7,800 back‑office positions betting that attrition plus AI would handle the load. Cyber‑security firm CrowdStrike shaved 5 % of its staff for the same reason. Even firms still growing keep whispering that their next cohort of junior analysts or HR coordinators might be the last big class.
Why the caution? Because AI shines at the boring bits—data entry, scheduling, basic report generation. Those tasks used to justify entry‑level jobs. Now a chatbot or document‑reading model can whip through them before your coffee cools. Companies don’t need to fire armies of assistants; they simply stop hiring new ones.
Illustrative number: the World Economic Forum’s Future of Jobs 2023 survey found 40 % of employers expect to trim staff where AI can automate tasks. That’s not every job, but it’s a hefty slice of white‑collar real estate.
Of course, doing things faster is useless if the output stinks. And wow, have we seen some stinkers.
Remember CNET’s experiment? The tech site quietly published dozens of AI‑written finance explainers. More than half had factual errors. Readers were not amused. A pair of New York lawyers got fined after ChatGPT invented case citations in a court brief—an expensive lesson in blind trust.
Customer service has its own horror stories. Klarna replaced 700 support agents with an AI bot in 2023. By early 2025 the CEO admitted the bot left shoppers fuming, and the company sprinted to re‑hire humans. Speed isn’t everything when refunds are on the line.
Yet the pendulum swings both ways. MIT researchers handed ChatGPT to a bunch of white‑collar volunteers and saw tasks finish 40 % faster with quality scores up 18 %. At a Fortune 500 call center, AI suggestions boosted rookie agents’ productivity by a jaw‑dropping 35 % while nudging customer‑satisfaction scores upward. The trick? Humans stayed in the loop. AI drafted, humans checked, customers smiled.
Lesson learned: treat AI like a sous‑chef, not the head chef. Let it chop onions at lightning speed, but taste the soup before it leaves the kitchen.
Platform‑economy thinker Sangeet Paul Choudary recently argued that the feel‑good slogan “AI won’t take your job, but someone using AI will” papers over deeper power shifts. When we stack his critique against the real‑world data above, a clearer picture emerges:
Tasks don’t just move; some disappear. IBM’s hiring freeze shows how whole swaths of back‑office work can be written out of the org chart once AI handles forms and approvals. That’s Choudary’s first fallacy in action: automation vs. augmentation isn’t a neat either/or—the task itself can lose value overnight.
Productivity isn’t the same as prosperity. The MIT and NBER studies prove AI can make individuals faster and even improve quality, but Choudary reminds us those gains don’t automatically flow to workers. If every analyst finishes ten decks before lunch, the going rate for deck‑making plummets while value concentrates in the hands of whoever owns the workflow (often the platform or the data‑rich firm).
Workflow rewrites create hidden fault lines. Klarna’s customer‑service fiasco and CNET’s error‑ridden articles illustrate what happens when firms rip out human checkpoints without redesigning accountability. Choudary calls this the “workflow isn’t sacred” fallacy: AI can skip steps or shift them elsewhere, sometimes dumping liability on the last human left in the chain.
Titles survive while bargaining power shrinks. Bookkeepers, junior lawyers, even graphic designers may keep their job labels, but as AI does more of the heavy lifting, pay premiums erode. The World Economic Forum’s job‑churn numbers hint at this quiet dilution: roles persist, yet more of the value moves upstream to AI‑rich hubs.
The net effect? Upskilling into “someone who uses AI” is necessary, but it’s only a life jacket in choppy waters. The bigger survival skill is reading how your organisation’s workflows and value pools are being redrawn—and positioning yourself where the new value accrues.
1. Admin & clerical
If your gig involves shifting information from one box to another, AI is already eyeing your chair. Data entry clerks, payroll processors, travel‑expense auditors—these roles top every “most automatable” list.
2. Customer service
Chatbots tackle FAQs 24/7 for pennies. Simple refund? AI’s got you. Tricky five‑step warranty claim? Still needs a human—at least for now. Result: fewer Tier‑1 agents, more specialists handling edge cases.
3. Finance & accounting
From automated loan underwriting to AI‑powered reconciliation, the number‑crunching side of finance is transforming. Bookkeeping clerks face the biggest risk; forensic accountants who can interpret AI red flags are doing fine.
4. Legal research
Large language models ingesting oceans of case law can spit out draft contracts in minutes. Junior associates who once spent nights proofreading clauses are suddenly supervising an algorithm instead. Verify everything—the model might hallucinate—but the throughput is wild.
5. Creative & design
Generative art shook illustrators in 2023 when Netflix Japan’s “The Dog & The Boy” used AI‑generated backgrounds. Agencies now pump out thirty ad mock‑ups with Midjourney before a human art director picks the winner. The market for average artwork is collapsing; demand for standout storytelling and brand voice is rising.
6. Software development
GitHub Copilot writes boilerplate code in seconds. Senior engineers love it; juniors worry. The best devs now architect systems and review AI output for edge‑case bugs instead of hammering out every for‑loop by hand.
(Notice the pattern? Routine in, AI out. Judgment and creativity still billable.)
Don’t grab your cardboard box just yet. Every wave of automation creates roles nobody imagined ten years earlier. AI is no different. Here are a few titles that barely existed pre‑2021:
The World Economic Forum forecasts a 30–40 % jump in demand for AI and data specialists by 2027. Somebody has to build, tune, and police these systems.
I’m not saying it’ll be painless. Transition periods rarely are. But history favors workers who partner with their machines instead of fighting them.
Will the robots take all our desks? Unlikely. They’re more likely to rearrange the office — shoving the gray, repetitive cubicles to one side while handing the humans better‑lit corners for creativity, empathy, and big‑picture thinking.
If we get the balance right, maybe Ana’s next new colleague won’t be another overwhelmed intern, but a tireless AI sidekick who handles the grunt work while she tackles projects that actually make her proud.
Yes, I used AI to help me properly structure this article.
Contents: