Start with the honest answer
Nobody can tell you exactly what your job will look like in the years ahead, and anyone who sounds certain, in either direction, is guessing. What you can reason about is how the change tends to arrive: less often as a job vanishing overnight, more often as a slow reshuffle of what fills your day.
A job is a bundle of tasks. Some of them are repetitive, follow clear rules and happen at a screen, and those are the ones AI tools are getting good at. Others involve people, trade-offs, context nobody ever wrote down, and responsibility for what happens next. Those change far more slowly, and in most jobs they're the part that really matters.
So rather than asking whether AI will take your job, it helps to ask two smaller questions. Which of my tasks could a tool do, or help with? And which parts of my work depend on me?
Think in tasks, not job titles
Take an ordinary week and write down what you actually spend your time on. Not your job description, the real thing: the emails, the meetings, the reports, the phone calls, the fixes nobody asked for. Then sort the list into three piles.
- Tasks a tool can already draft or speed up: first versions of routine documents, summaries of long email threads, meeting notes, reformatting, simple searches through information you already hold.
- Tasks where a tool helps but you stay in charge: analysis you then have to interpret, proposals you have to tailor, code or copy you have to check, research you have to weigh up.
- Tasks that rely on you: deciding what matters, calming an upset client, persuading a sceptical colleague, knowing which rule can bend and which can't, training someone new, being the person others trust to get it right.
What changes, and what that frees up
You may find the first pile is smaller than you feared and the last one is bigger than your CV suggests. The tasks in the first pile are also, quite often, the ones nobody enjoys: the copying, the formatting, the chasing of information that already exists somewhere.
When those shrink, the time doesn't vanish. It can move towards the work in the other two piles, if you steer it there. That's the real shift to watch for in your own role: less time producing first drafts, more time judging them; less time gathering information, more time deciding what to do with it; less time on routine, more time with people.
None of this is guaranteed, and some roles will change more than others. But the pattern gives you something to act on. The more of your week sits in the last pile, and the better others can see it, the stronger your position.
What stays human
AI tools are fluent, fast and tireless. What they lack is a stake in the outcome, a shared history with the people involved, and responsibility for what happens when things go wrong. That leaves a set of skills at the heart of almost every job, skills that are easy to undervalue precisely because they've never been written down.
- Judgement: weighing things that won't fit neatly in a spreadsheet, and making a call when the information is incomplete. A tool can list the options; someone still has to choose, and answer for the choice.
- Relationships: knowing that a client needs a phone call rather than an email, or that a colleague has gone quiet for a reason. That kind of attention comes from time spent with people.
- Trust: people act on advice from someone whose record they know. Trust is built by doing what you said you would, again and again, and it can't be generated on demand.
- Coordination: getting people with different priorities to move in the same direction, spotting the hand-over that's about to fail, keeping a project on course when the plan meets reality.
- Passing on know-how: showing a new starter how things really work here, which is rarely what's in the manual. Teams lean on the people who do this, and they're often the last to be credited for it.
Eight soft skills AI can't do for you
None of these shows up on a certificate, and none of them can be handed over to a tool. They're what people mean when they say someone is good to work with, and they're the part of your job that grows in value as the routine tasks get automated.
- Listening, so you understand what someone actually means, not just what they said.
- Staying calm, so you can think straight when everything speeds up.
- Explaining things simply, so other people can pick up what you know.
- Taking responsibility for a decision and for what follows from it.
- Curiosity, so you keep learning, including how to use new tools.
- Bringing people together, so different personalities pull in the same direction.
- Critical thinking, so you check what a tool suggests instead of copying it.
- Adaptability, so you can change how you work without losing the thread.
Spot your own soft skills
These are usually filed under soft skills, which makes them sound optional. They aren't. The trouble is that you're often the last person to notice your own: they feel like just the way you work, hardly worth mentioning. A few questions help bring them to the surface.
Write the answers down with an example next to each. A skill with a story attached is something you can talk about in an interview. A skill without one is just an adjective.
- What do colleagues come to you for, even when it isn't your job?
- When something goes wrong, which role do you end up playing: the calm one, the fixer, the one who rings the client, the one who works out what happened?
- What have you taught someone else, formally or not?
- Which situations do others avoid that you handle without much fuss?
- What would quietly fall through the cracks if you were off for a few weeks?
- What did your last good appraisal, or the last thank-you message you received, actually praise?
Get them vouched for by the people you've worked with
Soft skills have a proof problem. Anyone can write collaborative, adaptable and calm under pressure on a CV, and recruiters know it. What carries weight is the same claim coming from someone else: a manager, a colleague or a client who saw it happen and is willing to put their name to it.
That's why recommendations matter more, not less, as AI takes over more of the paperwork. A tool can polish a CV in no time. It can't produce a former colleague who describes, in their own words, how you held a difficult project together.
- Ask people who saw your work up close, rather than the most senior names you know.
- Ask them to describe a specific situation, not to list your qualities.
- Ask soon after you've worked together, while the details are still fresh.
- Ask a mix of people: someone above you, someone alongside you, someone you helped or served.
- Offer to write one back. How you describe other people says a good deal about you too.
Learn to use AI in your own work
You'll be in a far stronger position if you try the tools on your own work, early and on your own terms. The aim isn't to become an engineer. It's to know what the tools are good at, where they go wrong, and where they genuinely save you time, so the change happens with you rather than to you.
- Pick one tedious task you do every week and try handing the first draft to a tool. Check the result line by line.
- Notice where it gets things wrong. Knowing a tool's blind spots is a skill in itself, and a useful one.
- Follow your organisation's rules on what can and can't go into these tools, especially anything confidential or personal.
- Share what works with your team. Helping others adopt a new way of working is coordination and passing on know-how in action.
- Keep the judgement for yourself: you sign it off, and you answer for the result.
Make how you work visible, with Take Your Skills
Qualifications say what you've studied, and job titles say where you've been. Neither says how you work, and that's the part that's hardest to replace. It's worth making it visible on purpose: one line per role on your CV about how you worked and what came of it, a handful of short stories ready for interviews, and a running note of the thank-you messages you'd otherwise forget.
Take Your Skills exists for the part a CV can't prove. You share your profile link with the people you've worked with, and each of them writes a recommendation about what you're like to work with. If they're stuck for words, an AI assistant asks them a few questions and they answer in their own words, so the recommendation still sounds like them. AI helping people say what they saw, not inventing it.
Their recommendations sit together on a public profile in your name. You choose which ones are shown, and you can send the profile by link or download it as a PDF. From your public recommendations, Take Your Skills also writes a professional portrait of you: how the people you've worked with see you, rather than how you'd describe yourself.
- It's free, for you and for everyone who recommends you.
- People who write for you sign in with a code sent to their email address, so there's no password to invent.
- Add a personal email address and your recommendations stay with you, whichever job you're in.