In November 2025, a customer service operations manager named Jacques Reulet II was laid off. Not because he was bad at his job — but because the AI chatbots he'd been forced to manage for months had finally gotten "good enough" that his employer didn't need him anymore.
What's Actually Happening Right Now
Forget the vague predictions. Here's what real companies have actually done — with names, numbers, and outcomes.
The body count (so far)
| Company | What happened | Scale | Outcome |
|---|---|---|---|
| Klarna | Replaced customer service agents with AI chatbot | 700 agents eliminated | Service quality tanked. Now rehiring humans. |
| Chegg | ChatGPT destroyed demand for paid tutoring | 45% of workforce cut (388 people) | Stock in freefall. Company survival uncertain. |
| IBM | Automated HR and back-office functions | 8,000 jobs cut; 200 HR roles automated | CEO promised to hire grads, then laid off thousands. |
| Booking.com | Replaced customer service staff with AI chatbots | Undisclosed cuts | Customers report inability to reach humans for fraud issues. |
| Duolingo | Declared 'AI-first' strategy | Stopped using contractors for AI-doable work | Massive social media backlash. |
| UPS | Machine learning automated logistics roles | 20,000 workers laid off | CEO cited automation as enabling the cuts. |
The data behind the headlines
But "your job will transform" offers cold comfort when the transformation looks like "we no longer need 6 people to do what 2 people plus an AI can handle."
| Occupation Category | Projected US Shifts | Primary Driver |
|---|---|---|
| Office Support | ~3.7 million displaced | Automation of administrative tasks |
| Customer Service | ~2.4 million displaced | AI chatbots and self-service |
| Food Services | ~2.1 million displaced | E-commerce shift + automation |
| Sales (In-Person) | ~1.8 million displaced | E-commerce growth |
| Production Work | ~1.6 million displaced | Manufacturing automation |
AI job disruption is not a prediction — it's an ongoing event with named companies and published headcounts. The pattern: 30% of work hours face automation, but 170 million new jobs will be created globally (net +78M). The workers who understand where displacement is happening can position themselves on the growth side. The workers who don't are the next Jacques Reulet II.
The numbers are real. The question is personal: is YOUR job on the list? There's a 2-minute test for that.
The 4-Question Test: Is YOUR Job at Risk?
Question 1: Could someone describe your typical workday as a flowchart?
Question 2: Could you do your entire job without leaving your desk?
The irony: the remote work revolution proved that physical presence wasn't necessary for many jobs — and in doing so, proved those jobs could be done by AI too.
Question 3: Would a client/customer/patient care if you were replaced by a really good AI?
Question 4: Does your manager care about your creative judgment, or your throughput?
Your job title does not determine your AI risk — your daily tasks do. A "marketing manager" doing routine campaign execution faces completely different risk than one setting brand strategy and managing client relationships. Same title, different futures. The 4-question test reveals which future is yours.
Now that you know where you stand, here's what the risk landscape looks like across three tiers — starting with the roles that are being displaced right now.
Tier 1: Immediate High Risk (2026-2027)
Tier 1: Highest Automation Risk Jobs
Percentage of job tasks automatable by AI in the near term
1. Data Entry Clerks & Administrative Assistants
2. Customer Service Representatives (Tier 1)
This is where the Klarna story lives. Basic customer inquiries — order status, account information, password resets, standard troubleshooting — are being replaced by AI chatbots at scale.
3. Basic Content Writers (SEO, Product Descriptions)
Templated, formulaic content — product descriptions, basic SEO articles, listicles, standard social posts — falls squarely within generative AI capabilities.
4. Bookkeepers & Basic Accounting Clerks
Rules-based financial processing — transaction categorization, basic reconciliation, standard reporting — is ideal for AI automation. These are the flowchart jobs.
| Tier 1 Job | Automation % | AI Score | Timeline | Where displaced workers go |
|---|---|---|---|---|
| Data Entry | 90%+ | ~15/100 | Now-2027 | Data quality management, process automation oversight |
| Telemarketing | 90% | ~15/100 | Now-2027 | Complex B2B sales, relationship-based account management |
| Tier 1 Customer Service | 85% | ~25/100 | Now-2027 | Escalation handling, customer success, retention |
| Basic Content Writing | 80% | ~30/100 | Now-2027 | Content strategy, creative direction, original reporting |
| Bookkeeping | 85% | ~20/100 | 2026-2027 | Advisory accounting, complex tax planning, client-facing roles |
- Waiting for the layoff notice instead of transitioning proactively — by the time you're cut, the best positions are already filled
- Doubling down on the same skills ("I'll get even better at data entry") instead of pivoting to the human layer AI can't touch
- Assuming a different company will still need the same role — if AI automates bookkeeping at IBM, it automates it everywhere
- Ignoring the transition paths that don't require a degree — trade apprenticeships, certifications, and adjacent role moves are faster and cheaper
- Treating this as a personal failure instead of a structural shift — Chegg didn't fire people because they were bad at their jobs
Tier 1 displacement is not a forecast — it's a fact with a published body count. Data entry, basic customer service, templated content, and routine bookkeeping are being automated now. But every Tier 1 role has adjacent positions that remain protected: customer service → customer success, content production → content strategy, bookkeeping → advisory accounting. The exit path exists. The window for using it is closing.
Tier 1 is the immediate wave. But the next tier is in some ways more dangerous — because it targets people who think they're safe.
Tier 2: Near-Term Risk (2027-2028)
Tier 2: Near-Term Automation Risk Jobs
Percentage of job tasks facing automation pressure
5. Junior Software Developers
6. Paralegals & Legal Research Assistants
7. Financial Analysts (Basic Reporting)
8. Manual QA Testers
Test case creation, execution, and basic bug identification are increasingly automated. The manual tester clicking through test scripts is the data entry clerk of engineering.
Tier 2 is where Jacques Reulet II's warning matters most. These jobs are not vanishing — but the entry-level versions are, which means the path from junior to senior is getting shorter and narrower. The Career Ladder Collapse is the most underreported AI employment effect: companies don't announce "we're hiring fewer juniors" in press releases. They just do it. If you're early-career in any of these fields, developing senior-level judgment FAST is survival, not ambition.
Tiers 1 and 2 are the immediate and near-term waves. But there's a third tier with a longer fuse — and it targets some roles most people assume are untouchable.
Tier 3: Medium-Term Risk (2028-2030)
The pattern from Tiers 1 and 2 repeats here with a longer timeline: standard and routine versions of the role face pressure, while complex and relational versions remain protected.
| Occupation | Automation % | Why at risk | What stays human |
|---|---|---|---|
| Radiologists (Standard Imaging) | 60% | AI reads standard images as well as humans | Complex cases, patient context, procedural work |
| Non-Literary Translators | 70% | AI translation quality is near-professional | Cultural nuance, creative translation, high-stakes comms |
| Recruitment Coordinators | 65% | Resume screening is pattern matching | Relationship building, candidate experience, assessment |
| Loan Officers (Standard) | 55% | Standard underwriting is algorithmic | Complex cases, client relationships, exceptions |
| Tax Preparers (Basic Returns) | 65% | Standard returns are automatable now | Complex situations, planning, advisory |
Tier 3 roles face a 36-48 month transformation window. The Standard-Complex Split applies to every one of them: standard imaging vs. complex diagnostics, routine translation vs. high-stakes cultural interpretation, basic tax prep vs. strategic advisory. The role title stays the same. The daily work — and who gets to do it — changes completely.
The three tiers share one pattern. But there's a bigger structural truth underneath them — one that reverses decades of career advice.
The White-Collar Reversal
You were told: go to college, get an office job, you'll be safe. That advice is now backwards. And the data isn't subtle about it.
| Metric | Blue-Collar / Trades | White-Collar / Office |
|---|---|---|
| Goldman Sachs task automation % | 4–6% | 37–46% |
| Median AI Resistance Score | 91/100 | 68/100 |
| Brookings AI exposure level | Low | High |
| Frey & Osborne automation probability | 2.6% median | 18%+ median |
If your job involves sitting at a desk, processing information, and producing documents — you face higher AI risk than someone working with their hands in unpredictable environments. The White-Collar Reversal is the defining structural shift of AI employment: career resilience in the AI age comes from physical presence and human connection, not credentials and screens.
The White-Collar Reversal explains the structural pattern. But to see the pattern in action — the full cycle from hype to backfire to correction — there's one company story that tells it all.
The Klarna Lesson: Why "AI-First" Is Backfiring
In 2024, Klarna's CEO Sebastian Siemiatkowski declared the company would be "OpenAI's favorite guinea pig." They replaced 700 customer service agents with an AI chatbot. They froze hiring. They told the world this was the future.
By May 2025, the strategy was in reverse. Service quality had dropped measurably. Customers couldn't get real help for fraud and complex issues. The AI was great at simple queries — and terrible at everything that actually mattered for customer loyalty.
Klarna started rehiring humans.
- Phase 1: Hype — AI handles 80% of routine tasks. Executives see cost savings. They cut headcount aggressively.
- Phase 2: Degradation — The 20% of work that AI can't handle is the 20% that matters most. Customer satisfaction drops. Quality suffers. Edge cases pile up.
- Phase 3: Correction — Companies rehire humans — but fewer, for different roles. The new jobs are higher-skilled, higher-paid, and focused on the work AI can't do.
Booking.com is going through Phase 1 right now (replacing support staff). They'll likely hit Phase 2 within 12 months when customers can't resolve fraud through a chatbot. The playbook repeats.
The Klarna Cycle is a prediction machine: companies that go "AI-first" without understanding what AI cannot do will create demand for the exact human skills they just laid off. If you develop those skills now — complex problem-solving, relationship management, quality judgment — you'll be the person they hire back at a higher salary. The cycle's Phase 3 is the opportunity hiding inside the disruption.
The Klarna Cycle shows that AI-first strategies create their own correction. But what about the careers that aren't waiting for a correction — because AI is actively making them more valuable right now?
Jobs That Are Actually Growing Because of AI
| Growing Occupation | Median Pay | Growth | AI Score | Why growing |
|---|---|---|---|---|
| Nurse Practitioners | $129,210 | 40% | 93/100 | Aging population + physical care AI can't replace |
| Wind Turbine Techs | $62,580 | 50% | 89/100 | Energy transition + every site is physically unique |
| Mental Health Counselors | $59,190 | 17% | 97/100 | The human relationship IS the treatment |
| Electricians | $62,350 | 9% | 94/100 | Every job site is different. Every wire run is unique. |
| InfoSec Analysts | $124,910 | 29% | 64/100 | More AI systems = more attack surface = more security jobs |
| Physical Therapists | $99,710 | 14% | 89/100 | Physical presence + patient trust + hands-on treatment |
- Physical presence in unpredictable environments — electricians, plumbers, construction, skilled trades
- Deep human trust and emotional connection — therapists, counselors, nurses, social workers
- Managing the complexity AI creates — cybersecurity, AI operations, systems architecture
The first two are obvious. The third is the career opportunity most people miss.
AI doesn't just replace work — it creates new kinds of work. Every AI system needs humans who understand it, monitor it, fix it when it breaks, and bridge it with human needs. That's an entire economy emerging.
The top-tier AI-resistant occupations (scores 90+) have median BLS growth of 14.5% — nearly 5x the national average. AI is not just destroying jobs. It is creating massive demand in three zones: physical presence work that AI cannot do, human relationship work that AI cannot replicate, and AI management work that only exists because AI itself exists. The winners are not avoiding AI — they are positioning where AI drives demand upward.
The map is clear: some careers are shrinking, some are growing, and the split runs through the middle of most professions. The final question is the most personal one: what do you actually do about it?
What to Do If Your Job Is At Risk
If your role falls in Tier 1 or Tier 2, here's the plan. Not motivational platitudes — concrete moves with timelines.
Step 01: Know your timeline (and don't lie to yourself)
Step 02: Map your transferable human skills
List every skill you use daily. Now separate them into three categories:
- Category A: AI can do this — data processing, templated writing, routine analysis, standard procedures
- Category B: AI can assist but not replace this — complex problem-solving, stakeholder management, quality judgment
- Category C: AI cannot touch this — physical skills, emotional intelligence, trust-based relationships, creative vision
Step 03: Choose your direction: up, adjacent, or out
Step 04: Become the human-AI bridge (this is the real opportunity)
Step 05: Start now. Not when the layoff notice arrives.
Don't wait for displacement. Use your current employment as a launchpad:
- Take on projects that build your Category B and C skills
- Volunteer for AI implementation initiatives (you'll learn the technology AND prove your value)
- Start networking in your target direction — before you need a job, not when you need one desperately
- If your company offers AI training, take it. If they don't, invest in yourself. 90 days of deliberate skill-building changes your trajectory
The biggest mistake career changers make: waiting for "the perfect time." The perfect time was six months ago. The second-best time is this week.
An AI career transition has three directions: up (junior → senior within your field), adjacent (routine role → complex role nearby), or out (vulnerable field → structurally protected field). The Human-AI Bridge Strategy — using AI tools while providing judgment AI cannot — is the highest-value positioning regardless of which direction you choose. The cost of starting now is measured in effort. The cost of waiting is measured in leverage you'll never get back.
- 01AI displacement is happening NOW: Klarna (700 agents), Chegg (45% workforce), IBM (8,000 jobs), UPS (20,000 workers) — completed actions, not projections
- 0230% of US work hours could be automated by 2030, but 170M new jobs will be created globally (net +78M) — this is transformation, not extinction
- 03The Standard-Complex Split: in every profession, routine/standard tasks face automation while complex/relational tasks remain protected or grow
- 04Tier 1 (data entry, basic customer service, routine content) faces 12-24 month displacement. Tier 2 (junior devs, paralegals, basic analysts) transforms over 24-36 months
- 05The White-Collar Reversal: AI threatens office workers (46% automatable) more than trades (4-6%) — a licensed electrician is more structurally protected than a paralegal
- 06The Klarna Cycle: companies that go 'AI-first' discover AI can't handle the 20% of work that matters most — then rehire for higher-skilled, higher-paid roles
- 07The Career Ladder Collapse: AI's most insidious effect is eliminating entry-level positions, narrowing the path from junior to senior
- 08The winning move: become the Human-AI Bridge — use AI tools while providing judgment, relationships, and physical presence that AI fundamentally cannot
Will AI replace all white-collar jobs?
No — but AI will transform white-collar work more dramatically than blue-collar work, reversing decades of automation patterns. The White-Collar Reversal shows white-collar roles averaging 68/100 on AI resistance vs. 91/100 for trades. Roles requiring complex judgment, human relationships, and accountability remain protected. Roles involving routine information processing are most vulnerable.
How accurate are these automation predictions?
These projections synthesize data from McKinsey, WEF, Goldman Sachs, and Frey & Osborne — among the most rigorous available. Our AI Resistance Scores correlate at r = −0.81 with established automation probability models. However, all predictions involve uncertainty — the Klarna Cycle proves that displacement timelines are not linear. The directional trend matters more than specific percentages.
Should I avoid learning to code because AI will replace programmers?
Absolutely not. Understanding code and AI systems makes you MORE valuable, not less. AI replaces routine coding — but developers who architect solutions, manage stakeholders, AND leverage AI tools are among the highest-paid professionals in the market. The role transforms dramatically but does not vanish. Learn to code AND learn to direct AI. That combination is the Human-AI Bridge in action.
Is it too late to transition if my job is in Tier 1?
It is not too late, but the window is narrowing. Tier 1 workers should begin active transition planning immediately. Many skills transfer directly to protected roles, and several AI-proof careers do not require college degrees: trade apprenticeships are paid training, home health aide certification takes weeks, customer success roles value service experience. See our AI-Proof Jobs Guide for specific entry paths from every starting point.
What if I can't afford to go back to school?
Many of the strongest transition paths require no formal education. Skilled trades use paid apprenticeships (earn while you learn — zero tuition). Home health aides need only weeks of certification. Moving within your field (customer service to customer success, content production to content strategy) requires demonstrated skills, not new credentials. The most expensive mistake is not investing in training — it's waiting until unemployment forces the decision.
Are remote jobs more at risk than in-person jobs?
Yes — and the data is striking. Physical Presence is the strongest single predictor of AI resistance in our model (r = −0.74). Remote work proved that physical presence was not necessary for many knowledge jobs. That same proof of concept applies to AI: if a job can be done from anywhere by anyone, it can eventually be done by AI from nowhere. Roles requiring in-person presence in variable environments are structurally more protected.
Will AI create more jobs than it destroys?
The WEF projects 170 million new jobs created vs. 92 million displaced by 2030 — a net gain of 78 million globally. McKinsey estimates $2.9 trillion in economic value from human-AI collaboration. New categories include AI operations, prompt engineering, AI ethics, human-AI interface design, and expanded roles in healthcare and trades. But 'more jobs created' does not help you personally if the new jobs require skills you don't have. That's why transition planning matters now, not later.
Prepared by Careery Team
Researching Job Market & Building AI Tools for careerists · since December 2020
- 01Jobs AI Can't Replace: AI Resistance Scoring for 30 Occupations (Original Research) — Careery Research (2026)
- 02The Future of Employment: How Susceptible Are Jobs to Computerisation? — Carl Benedikt Frey & Michael A. Osborne, Oxford Martin School (2013 (updated 2017))
- 03How Will AI Affect the Global Workforce? — Goldman Sachs Research (2023–2024)
- 04Generative AI and the Future of Work in America — McKinsey Global Institute (2023)
- 05Agents, Robots, and Us: Skill Partnerships in the Age of AI — McKinsey Global Institute (2025)
- 06The Future of Jobs Report 2025 — World Economic Forum (2025)
- 07Occupational Outlook Handbook (2024–2034 Projections) — U.S. Bureau of Labor Statistics (2025)
- 08What Jobs Are Affected by AI? Better-Paid, Better-Educated Workers Face the Most Exposure — Brookings Institution (2023)
- 09Going 'AI First' Backfires on Duolingo and Klarna — Fast Company (2025)
- 10Klarna CEO Says AI Helped Company Shrink Workforce by 40% — CNBC (2025)
- 11Chegg Slashes 45% of Workforce, Blames 'New Realities of AI' — CNBC (2025)
- 12IBM Cuts 8,000 Jobs Amid AI Expansion — Computing.co.uk (2025)
- 13Companies Are Replacing Workers With AI — Fast — Forbes (2025)