AI and jobs 2026: jobs disappearing and emerging

In one sentence

AI does not replace humans : it replaces humans who don't use it. In 2026, 35% of tasks in offices are automatable by AI, but only 8% of jobs will disappear completely. The real revolution: every profession changes profoundly, and new AI-native jobs are emerging at high speed.

💼

The analogy that works

AI is to the office worker what the tractor was to the farmer in the 20th century. The tractor didn't kill farmers, it massively reduced their number (from 50% of the active population in 1900 to 2% today), whilst increasing agricultural production by 10x. Same with AI: fewer white-collar workers, but far more productive individually. Question: will you be the one driving the tractor or the one it replaces?

🤖 Want to know how AI really works?

Before being afraid, understand what it can and can't do.

Read the LLM article

The numbers that hurt (but also reassure)

AI impact on jobs (2025-2026 studies)

% workers using AI daily (France)47%

% companies having created AI positions (2024-2026)41%

% jobs heavily transformed35%

Average productivity gain for knowledge workers28%

% companies having cut positions via AI19%

% new jobs created thanks to AI12%

% jobs TOTALLY replaced (OECD 2026)8%

Honest reading: yes, 8% of jobs disappear (1.5M in France), that's massive and brutal. But 12% are created (2.3M) in parallel. And 35% are deeply transformed. The real challenge: the transition.

Map of jobs facing AI in 2026

The 4 zones of jobs facing AI

According to level of automatability and creativity required. Know where you stand.

The 6 jobs that are DISAPPEARING (or heavily shrinking)

📚Jobs at risk: what's really happening

1. 📞 Level 1 call centre operators

2026 status: -45% positions vs 2022. GPT-5 + voice AI handle 80% of simple calls (enquiries, basic complaints).

What remains: level 2-3 (complex cases, emotional de-escalation, negotiations).

Action: upskill to "product specialist" or shift to QA/coaching.

2. 📋 Data entry

2026 status: -60% positions. OCR + LLM read invoices, forms, CVs automatically.

What remains: complex data entry (multi-source, critical quality).

Action: become a data analyst (analyse, don't input).

3. 📝 Low-end SEO writers

2026 status: -70% of "£0.01/word web writer" positions. GPT-5 does better for £0.

What remains: expert writers, investigative journalists, high-level copywriters.

Action: move upmarket (sector expertise, unique voice, interviews, original data).

4. 🌐 Generalist translators

2026 status: DeepL + Claude = 95% quality of a human translator. -50% positions.

What remains: certified translators, literary, highly specialised technical (medical, legal, patents).

Action: extreme specialisation or shift to revision/post-editing (proofread the AI).

5. 🧾 Basic accounting

2026 status: -40% "junior accountant" positions. AI classifies, reconciles, generates statements.

What remains: tax advice, auditing, financial strategy.

Action: become a specialist chartered accountant or management consultant.

6. 🎨 Junior UI designers / commercial illustrators

2026 status: Midjourney v8 + Figma AI = 80% of design tasks.

What remains: art directors, strategic designers, illustrators with a strong voice.

Action: become AD or specialise in a unique artistic niche.

The 7 jobs that are EMERGING (and pay well)

AI jobs recruiting in 2026

	💼Job	💰Salary France gross/year
AI Engineer / Prompt Engineer	Creates and optimises AI solutions for businesses	£48-105K
AI Risk Manager / AI Auditor	Assesses AI risks (AI Act compliance, security)	£52-96K
MLOps Engineer	Deploys and maintains models in production	£52-114K
AI Trainer / RLHF Specialist	Trains and fine-tunes AI models	£44-79K
AI Product Manager	Manages products that integrate AI	£61-131K
AI Ethicist / Policy	Ethical framework, regulatory compliance	£48-88K
Data Engineer (RAG/Vector)	Architects data for AI	£48-96K

Reading: these jobs pay 30-50% more than equivalent traditional jobs. Trend that will continue for 5-10 years.

Timeline of transformation 2024-2030

Timeline of job disruption by AI

When each category of job will be massively impacted

How to protect yourself: the 5-point strategy

🛡️

Becoming AI-native in 12 months

1. Assess your risk level honestly Look at the map above. Be honest: is your job routine? If yes to 50%+, start moving now, not in 2 years. 2. Learn to use AI in your job Don't stop at "ChatGPT to summarise". Learn the pro tools of your field: - Marketing → Jasper, Copy.ai, Make - Code → Cursor, Claude Code, Copilot - Design → Figma AI, Midjourney, Runway - Legal → Harvey, Spellbook - Accounting → Pennylane (with AI), Indy 3. Move up a notch The junior does tasks (which AI can do). The senior does strategic advice (which AI can't). Aim for +1 level in your job. 4. Build your anti-AI "moats": - Human relationships: clients who know you - Rare expertise: niche domain that few master - Taste and judgement: ability to judge what's good - Accountability: signing/endorsing (AI can't) 5. Train 3-5h/week on AI - 1 online course (Coursera, Udemy) - 1 AI newsletter (TLDR AI, Snacks AI, nAIvigate) - 1 tool tested each week

Is the "doomer" scenario true?

The AGI risk: should we worry?

Some experts (Geoffrey Hinton, Yoshua Bengio) predict an AGI (Artificial General Intelligence) between 2027 and 2035. An AGI would do all cognitive jobs better than a human. The doomer scenario: "All white-collar jobs disappear in 5 years." The 2026 reality: - ✅ AGI doesn't exist yet (LLMs have profound limitations) - ✅ Technology diffusion takes 10-30 years typically (cf. Internet, smartphones) - ✅ Regulations slow it down (AI Act, etc.) - ⚠️ But not preparing = playing roulette No-regret strategy: act as if AGI arrives in 2030, but invest in skills that are valuable even if it arrives in 2050. Soft skills, creativity, leadership, rare expertise = useful in all scenarios.

What's reassuring (even doomers say it)

📚Why AI probably won't kill all jobs

1. Polanyi's paradox

"We know more than we can tell." Many tasks seem simple but are impossible to formalise (e.g. discerning a good dish, understanding a capricious client). AI can approximate, but remains imperfect.

2. Jevons paradox

When a service becomes cheaper (AI productivity), demand explodes. The more we automate code, the more software we create. More software = need for more devs (but different devs).

3. Humanity prefers humans in certain contexts

Healthcare (patients prefer a human)
Education (teacher-student relationship)
Strategic advice (legal + human accountability)
Complex B2B sales
High-end creativity (artists, brands)

4. AI creates more wealth → more needs

More productivity → more income → more demand for: tourism, healthcare, leisure, experiences, sport, art. These sectors are exploding in 2026.

5. Regulation will slow disruption

The AI Act (EU), SB-1047 (USA), sector constraints (healthcare, finance, legal) force keeping a human in the loop. Not out of sentimentalism: out of legal accountability.

The metaphor that sums it all up

🚂

The shift from animal traction to motors

In 1900, 30% of jobs in the USA were horse-related (coachman, groom, carriage maker, saddler, horseshoe maker, stable, hay...). In 1930, 2%. These 28% didn't disappear unemployed. They became: - Auto mechanics (job that didn't exist) - Truck drivers (job that didn't exist) - Petrol station attendants (job that didn't exist) - Car salespeople (job that didn't exist) - Road/motorway designers (job that didn't exist) → The transition was HARD for individuals (loss of income for 10-20 years), but society globally progressed. The 2026-2040 challenge: managing the human transition, not the technology. If you're in a disrupted job, you have 5-10 years to switch. Don't waste 3 years waiting.

Remember absolutely

✅ 8% of jobs will disappear completely (1.5M in France)
✅ 35% of jobs will be heavily transformed
✅ 12% of NEW AI jobs are being created in parallel
✅ At-risk jobs: routine, non-creative (call centres, data entry, SEO writing)
✅ Winning jobs: creative + relational + expert (doctors, senior designers, executives, manual trades)
✅ Your strategy: use AI + move up a notch + build your "moats" + train 3-5h/week
✅ Don't panic: the transition takes 10-20 years, you have time if you start now

AI will massively transform jobs, but it won't kill the majority of them before 2035. The real question: are you moving, or are you waiting?

🧠 Quiz

Question 1 of 3

What percentage of jobs will really DISAPPEAR completely because of AI according to OECD 2026 studies?

Going further

🦜 What's an LLM?, the tech behind disruption
💻 Vibe coding: coding with AI, the typical example of job evolution
🔍 GEO: AI-version SEO, how marketing evolves

AI and jobs: which professions disappear and emerge in 2026?