AI and the Job Market: 92 Million Jobs at Risk — or New Opportunities?

Across Europe, tens of thousands of positions fell to AI automation in 2025. SAP announced the restructuring of 8,000 roles, Klarna in Stockholm cut most of its customer service team. In Germany, Austria, and Switzerland, companies are feeling the shift acutely. Behind every number is a person who woke up one morning to learn that a machine now does their job faster and cheaper.

Let's talk honestly about what's happening. Not with fear-mongering, not with naive optimism. But with a clear look at the facts — and at the people they affect.

The World Economic Forum predicts: 92 million jobs will be displaced by AI and automation by 2030. At the same time, 170 million new ones will be created. A net gain of 78 million. Sounds reassuring, right?

The problem with net figures: they obscure reality. When a 55-year-old administrator in Munich loses their job to AI and the "new jobs" are AI developers in Berlin or London, that's not an exchange — it's a rupture. The macro statistics show a gain; the micro level shows lives falling apart. Germany's IAB (Institute for Employment Research) warns that the DACH region, with its strong Mittelstand, faces particularly significant adjustment challenges.

McKinsey estimates the share of automatable work hours at 60 to 70 percent of all activities. This doesn't mean 70 percent of jobs will disappear — but it means nearly every job will fundamentally change. The question is no longer "Am I affected?" but "How much?"

Those affected are not, as often assumed, factory workers or low-skilled laborers. The current AI wave primarily hits cognitive routine work — and thus the middle of society.

Customer service tops the list. Chatbots and AI agents can now answer 80 percent of typical inquiries — faster, around the clock, in any language. Klarna replaced the majority of its customer service team with AI in 2024 and reported better customer satisfaction. That's the bitter punchline: the machine doesn't just do it cheaper, it partly does it better.

Content creation is being massively disrupted. SEO texts, product descriptions, social media posts — for standard content, hardly anyone needs a copywriter anymore. The content mill industry, which lived on cheap freelancers for years, is being replaced by AI that's even cheaper.

Translation and localization was one of the first areas hit. DeepL and others have already transformed the market for simple translations. Literary translation remains human — but that's a niche market.

Data analysis and reporting is changing fundamentally. What junior analysts used to spend days on — preparing data, identifying patterns, writing reports — an AI handles in minutes. The entry-level positions through which young analysts used to enter the industry are vanishing.

The common thread: structured, repeatable cognitive work. Everything that follows a pattern is automatable. And AI keeps getting better at recognizing patterns that humans never perceived as patterns.

There's a particular cruelty in having a machine take over creative work — trained on that very creative work. This isn't an abstract consideration. It's the reality of millions of creatives worldwide.

The Writers Guild Strike 2023 lasted 148 days and brought Hollywood to a standstill. The screenwriters weren't fighting low wages — they were fighting their own obsolescence. Studios had begun using ChatGPT for initial script drafts and employing human writers only for revisions. At half the price.

Visual artists face an existential crisis. Midjourney, DALL-E, and Stable Diffusion can generate images in seconds that would take an illustrator hours or days. That these models were trained on the works of those very artists makes it especially bitter. Your style gets replicated, and you're not even asked — let alone paid.

In the music industry, the fight has only just begun. Suno and Udio can generate songs in any style — including voices that sound uncannily like real artists. Universal Music, Warner, and Sony have filed lawsuits. The message is clear: the creative industry won't go down without a fight.

In the debate about AI and jobs, a crucial perspective is often lost: there are capabilities AI doesn't have — and that can't simply be programmed.

Complex judgment: AI can recognize patterns, but it cannot judge whether a decision is right — only whether it's statistically probable. A judge deciding an individual case, a doctor interpreting symptoms in the context of a life story, a manager reading the mood of their team — these require judgment, not pattern recognition.

Empathy and emotional intelligence: Chatbots can sound empathetic. But between simulating and feeling empathy lies a difference every person can sense. Care, therapy, crisis counseling, grief support — wherever human connection is essential, AI remains a tool, not a replacement.

Original creativity: AI can remix, combine, vary — but it cannot create the unexpected that emerges when a person draws from personal experience, cultural context, and emotional depth. An AI-generated image can be pretty. Art is something else.

Relationship work: Negotiations, leadership, mentoring, networking — everything based on trust, personality, and relationship history remains deeply human.

Ethical judgment: Not "What is statistically optimal?" but "What is morally right?" — a question AI fundamentally cannot answer because it has no moral consciousness.

The most productive perspective on AI in the workplace is not replacement but augmentation. The copilot model: AI handles the tedious parts, humans make the decisions.

Concrete examples of how this works in practice:

• An AI analyzes 10,000 customer feedbacks and identifies patterns — an analyst interprets the results and derives recommendations.
• An AI creates an initial draft of a report — an expert reviews it, adds context, and turns it into a decision-ready document.
• An AI filters 500 applications and identifies the most promising — a recruiter conducts the interviews and makes the hiring decision.
• An AI translates a contract — a lawyer reviews the nuances and cultural contexts.

In each case, the work gets better, not less. The entry barrier drops, quality rises, and humans can focus on what they do best: judging, deciding, taking responsibility.

But this also means: job profiles are fundamentally changing. An analyst who can't use AI tools will dramatically lose productivity compared to one who can. The skill "working with AI" is becoming as fundamental as "working with Excel" was 20 years ago.

The responsibility for managing this transition doesn't lie with individual workers. Anyone who believes that is making it too easy.

Companies have a duty to prepare their employees for change — before replacing them. This means:

1. Retraining programs that aren't PowerPoint courses but genuine qualification for new roles.
2. Transition periods instead of immediate layoffs — when a department is automated, affected employees need time and support for the transition.
3. Transparency: when AI deployment is planned, employees must be informed early — not when the termination letter arrives.
4. Sharing productivity gains: when AI boosts productivity by 40 percent, the benefits shouldn't flow exclusively to shareholders.

Policymakers must create the framework:

• Social safety systems that cushion the transition — not everyone can retrain themselves.
• Education policy that integrates AI competency into curricula — from elementary school to university.
• Regulation that obliges companies to deploy AI responsibly — the EU AI Act is a start, but it only marginally addresses labor market effects.
• Support for industries and regions particularly affected — structural change requires structural support.

The question "Will AI take my job?" is the wrong question. The right one is: "How will AI change my job — and am I prepared for it?"

The honest answer: some jobs will disappear. There's no sugarcoating that. 92 million is an abstract number — until it's your job. But new roles will also emerge that don't exist yet. Ten years ago, there were no prompt engineers, no AI ethicists, no data storytellers.

What matters is how we shape the transition. Adaptation requires support, not platitudes about "lifelong learning." Companies must take responsibility for the people whose work they automate. Policymakers must create safety nets that make the transition bearable. And each of us must be willing to learn new skills — even when it's uncomfortable.

Those who adapt will thrive. But adaptation must not be a matter of individual luck. It must be a societal mission.

deepsight follows the augmentation model: AI text analysis that complements human expertise, not replaces it. Learn more.