Few questions cause as much anxiety as “is AI going to take my job?”. By 2026 there’s enough real-world experience to answer without doom-mongering and without denial. Here’s what’s genuinely changing in employment, what’s noise, and above all what you can do, whatever your job is.
What is genuinely changing
- Tasks are transforming, more than whole jobs: AI absorbs specific parts of a role (drafting, summarising, classifying, finding information), not the entire role. Work changes shape before it disappears.
- Repetitive, digital work is the most exposed: the more predictable and text- or data-based a task is, the sooner AI touches it. Physical work, relationships and anything requiring contextual judgement resist far more.
- New tasks are appearing: supervising, correcting and directing AI is already part of daily life in many roles. Someone has to decide what to ask it, evaluate what it returns and take responsibility for the result.
- The gap is about adoption, not age or industry: on the same team you’ll find someone multiplying their pace with AI and someone working exactly as before. That difference is starting to show in who gets which projects.
Our take: less apocalypse, more homework
- Distrust round numbers. Every study gives different percentages of “jobs affected” because they measure different things. Our rule: if a headline scares you with an exact figure, check who published it and what they’re selling.
- The real short-term risk isn’t “AI replaces you”, it’s being left out of the conversations where how work gets done from now on is decided. Those who understand the tool take part; those who don’t, endure it.
- What using AI daily has taught us: it does the draft very well and the judgement very badly. Everything that involves deciding, prioritising, dealing with people and taking responsibility remains human ground. That’s where it pays to grow.
Our honest opinion: the panic narrative paralyses, and so does the “nothing will change” one. The useful spot is in the middle: your job will almost certainly change; whether it changes in your favour depends quite a lot on what you do these years.
How to adapt, in practice
- Use AI in your own work now: pick a task that bores you (reports, emails, summaries) and solve it with AI for a month. You learn by using it, not by reading about it.
- Make visible what AI doesn’t do of you: judgement, client relationships, knowledge of context, accountability. Strengthen it and make it noticeable.
- Learn the basics well: what a prompt is, what a hallucination is, what a model can and can’t do. Our route to learn AI from scratch is more than enough to start.
- If you lead a team, set clear rules: what can be done with AI, what gets reviewed and who answers for the result. Productivity without judgement gets expensive.
- Don’t train in a panic: an expensive course doesn’t protect you more than using the tool every day. Consistency beats intensity.
Frequently asked questions
Which jobs are most at risk from AI?
More than whole jobs, tasks: the repetitive, predictable, text- or data-based ones. A single role can have half its tasks highly exposed and the other half nearly untouchable.
Is it worth retraining towards AI?
Learning to use it well, yes, whatever your field: it’s transversal, like office software was. Switching careers to “something AI” out of fear alone, no: do it if it also genuinely interests you.
Conclusion
AI is changing employment task by task, not overnight. The best defence is boring but it works: use it daily, strengthen what it can’t do and don’t make decisions out of panic. To get moving, start with the route to learn AI from scratch and understand the real state of AI agents, which are what will move the board the most.
