Is AI Really Killing Jobs? What Experts Actually Say
You're scrolling through LinkedIn and the headlines keep coming: Microsoft lays off 19,000 people, Intel cuts 35,500 jobs, Amazon slashes 14,000 positions. The usual suspect? AI. And you're wondering if your own job is next. Spoiler alert: the reality is far more nuanced than the alarmist headlines suggest. I've dug through the studies, compared the numbers, and consulted the experts to give you the real picture as we close out 2025.
In this article:
1. The scary numbers (and why they're misleading)
Honestly, when you look at the raw stats, it's pretty alarming. 94,000 tech jobs cut in 6 months in the US alone, according to Challenger's June 2025 report. That's over 400 positions per day. Microsoft, IBM, Meta, Intel... the giants are falling one after another.
The thing is, these numbers need to be unpacked. When Intel announces 35,500 layoffs, it's less about AI than a massive restructuring to catch up with NVIDIA. When Amazon mentions 14,000 positions, CEO Andy Jassy himself said: "This decision isn't really driven by AI, at least not yet." He points instead to excessive bureaucracy.
According to Challenger, Gray & Christmas, AI was cited as a reason for 48,414 job cuts in the US since the beginning of 2025. That represents about 20% of total layoffs. A lot? Yes. But not the apocalypse we're being sold.
The problem is that companies are using AI as a convenient excuse. It sounds better to investors to say "we're optimizing with AI" than "we mismanaged our headcount during the post-Covid boom." SAP's CFO put it bluntly: "For the same production volume, we can afford to have fewer staff." At least that's honest.
Key figures to remember:
- US tech job cuts (H1 2025): 94,000
- Layoffs citing AI: 48,414 (~20%)
- Companies planning AI-related cuts by 2030: 41%
- Global jobs at risk according to Goldman Sachs: 300 million
- Jobs that will be "impacted" according to IMF: 40%
2. The other side of the coin: job creation
OK, now let's take a breath and look at the other side of the picture. Because while headlines scream about the apocalypse, there's some serious job creation happening.
In France, 166,000 AI-related job postings were published in 2024 according to PwC's AI Jobs Barometer. That's ahead of Germany (147,000) and the UK (125,000). And jobs "augmented" by AI grew by +252% between 2019 and 2024.
The World Economic Forum predicts 92 million jobs will be eliminated by 2030... but also 78 million created. The net balance is negative, sure, but we're far from a doomsday scenario where everyone ends up unemployed.
The IBM case is fascinating. They laid off 8,000 HR employees replaced by their AskHR chatbot (which handles 94% of requests). But then they hired massively in software engineering and marketing. CEO Arvind Krishna explains: "Our total headcount has increased thanks to investments in high-value sectors."
The real game changer is the AI salary premium. Professionals trained in AI earn on average 56% more than their peers without these skills. In France, a junior AI position starts between 45-70k euros, while seniors easily exceed 100k euros.
3. What scientific studies actually say
I spent quite a bit of time digging through serious studies (not LinkedIn polls). And the verdict is... mixed. As often in science, the answer is "it depends."
The Yale University study published in October 2025 made waves. Their conclusion: "AI has not had a significant effect on employment so far." The researchers explain that major technological transformations historically unfold over decades, not months.
A Vanguard study from December 2025 reaches the same conclusion: jobs most exposed to automation showed employment growth of 1.7% between mid-2023 and mid-2025. That's higher than pre-pandemic levels (1%). Counterintuitive, right?
But there are also concerning signals. The Stanford study from August 2025 (Brynjolfsson, Chandar, and Chen) shows a real difference depending on how companies use AI. Those that "augment" human work hire more. Those that see AI as a substitute hire less. Makes sense.
The IMF predicts that 40% of global jobs will be affected by AI, with a stronger impact in wealthy countries (60%) than in emerging economies (40%). And in most scenarios, AI will worsen inequality. Not great.
| Source | Conclusion | Nuance |
|---|---|---|
| Yale (Oct 2025) | No measurable impact | Divergence for juniors |
| Vanguard (Dec 2025) | Exposed jobs growing | Local adjustments possible |
| Stanford (Aug 2025) | Impact depends on usage | Substitution vs augmentation |
| IMF (2025) | 40% jobs impacted | Increased inequality |
| Goldman Sachs | 300M exposed | Exposure does not equal elimination |
| PwC (2025) | AI redefines, doesn't eliminate | Productivity x3 in 6 years |
4. Jobs at risk vs jobs that are booming
The thing is, AI doesn't affect everyone equally. There are clear winners and losers in this story. And spoiler: it's not always who you'd expect.
The most exposed jobs are those with repetitive, standardized tasks: administrative assistants, data entry operators, call center agents, tier-1 customer support. At IBM, their AskHR chatbot absorbed 94% of traditional HR functions. At Microsoft, 30% of code is now generated by GitHub Copilot, reducing the need for junior developers.
But on the flip side, some jobs are literally exploding. Data scientists, prompt engineers, AI ethics specialists, cloud architects... In France, 751 AI startups already employ 36,000 people, and 92% of them want to hire more.
+ Jobs that are booming
- - Data Scientist - Average salary 4,500 euros/month
- - Prompt Engineer - Highly sought-after new role
- - AI/ML Developer - Junior 45-70k euros
- - AI Ethics Officer - Strategic role
- - Cloud Architect - Talent shortage
- - GenAI Consultant - 40-55k euros junior, 80k+ senior
- Jobs under pressure
- - Tier-1 Customer Support - Chatbots taking over
- - Data Entry - Massive automation
- - Administrative Assistants - Repetitive tasks targeted
- - Junior Developers - Copilot generates 30% of code
- - Basic Quality Control - AI vision more precise
My Advice
Stop asking yourself if AI will take your job. The real question is: will someone who masters AI take your job? That's a huge difference. In 2025, the biggest risk isn't being replaced by AI, but being replaced by a human who knows how to use it. Start by training on the tools in your industry. Even 30 minutes a day on ChatGPT, Claude, or tools specific to your field is already an investment that pays off.
5. How to prepare yourself (without panicking)
OK, now that we've looked at the situation realistically, let's talk practicalities. Whether you're employed, freelancing, or job hunting, here's what you can do right now.
Free resources worth checking out
Microsoft launched its AI Skills Plan which has already reached 700,000 people in France. Their AI Skills Navigator platform is free and accessible to everyone. The goal: train 1 million French people by 2027. Google also offers its Digital Workshops with a specialized AI curriculum. And Microsoft AI Schools by Simplon are free and open without degree requirements.
Available funding options
In France, you have several options to fund your upskilling: the CPF (Personal Training Account), the CEP (Career Development Counseling) which is free, the PTP (Professional Transition Project) for longer career changes, and the AIF (Individual Training Assistance) if you're a job seeker.
Priority skills to develop
According to recent studies, here are the most in-demand skills: Python and SQL for technical basics, Power BI and Tableau for data visualization, Machine Learning fundamentals, and especially the ability to collaborate with generative AI tools. 81% of workers believe that with the right training, it's entirely possible to learn to use AI, even without prior knowledge.
Free resources to get trained:
- Microsoft AI Skills Navigator - skills.microsoft.com
- Google Digital Workshops - grow.google
- Coursera AI Courses - coursera.org
- DeepLearning.AI - deeplearning.ai (free introductory courses)
- edX AI Programs - edx.org
- LinkedIn Learning - linkedin.com/learning
Frequently Asked Questions
Will AI really eliminate my job?
It depends on your profession and how your company uses AI. Studies show that only 5% of positions in France are fully automatable. Most jobs will be transformed, not eliminated. The real risk is not training yourself and being overtaken by colleagues who master these tools.
How long does it take to retrain in AI?
It can range from a few weeks for basic certification to 1-2 years for a full master's degree. Intensive bootcamps like those from General Assembly or Springboard allow you to acquire operational skills in 3-6 months. The most important thing is to start somewhere, even with 30 minutes a day.
What are the highest-paying AI jobs?
In Europe, senior AI engineers exceed 100k euros annually. Data scientists average around 50-70k euros. In the US, salaries climb to $120-200k for senior profiles. Professionals trained in AI earn on average 56% more than their peers.
Are government AI training initiatives effective?
Various countries are launching AI training programs with varying levels of success. France's "Embrace AI" plan launched in July 2025 aims to train 15 million people by 2030 with a 200 million euro budget over 5 years. It's a start, but many experts believe it's insufficient given the scale of the challenge. Private initiatives from Microsoft and Google complement these efforts.
Conclusion
So, is AI really killing jobs? The honest answer is: yes and no. Yes, some jobs will disappear or be radically transformed. No, it's not the apocalypse some are selling us.
The numbers show a nuanced reality: 94,000 tech jobs cut, but 166,000 AI positions created in France alone. 41% of companies plan cuts, but 78 million jobs will be created by 2030. The real issue isn't the disappearance of work, but its profound transformation.
My advice? Stop stressing about headlines and start taking action. Train yourself, even in small doses. Free resources exist. And remember: 60% of workers will need retraining by 2030. Might as well get ahead of the curve.