AI Guides › Career & Work

What To Actually Say When Asked "How Do You Use AI At Work?"

By Nigel Guy · 2 min read

This question shows up in interviews, performance reviews, and casual conversation with a new manager now, and most people answer it in one of two equally unconvincing ways: a vague "yes, all the time, it's great" that says nothing, or a defensive "I use it a little but I always check it carefully" that sounds like an apology. Neither answer is a lie, exactly — they're just answers with no actual information in them, delivered by someone who hasn't thought about the question until it was asked.

The mechanism: the three-part answer

1. Name one real task, specifically. Not "AI" as a category — "I use it to draft the first pass of a weekly summary" or "to get a rough structure going before I write the actual thing." One concrete example beats a general claim every time.
2. Say what role it played, honestly. Did it draft, did it check your work, did it generate options you picked from? Naming the actual role is more convincing than "I use it a lot," and it's the part most people skip.
3. Name one thing you specifically check or wouldn't delegate. This is the part that turns the answer from a slogan into a judgement — "I don't let it write anything that goes to a client without reading it myself first" tells the asker something real about how you think.
4. Have a second example ready, not just the first one. One example can sound rehearsed; a second, different one signals it's actually a habit rather than a single anecdote you prepared.
5. Say it plainly, without over-selling or over-apologising. This question usually isn't a test of your enthusiasm or your caution — it's a test of whether you've actually thought about your own working habits at all.

What to skip

Skip listing tool names as if the brand is the point — nobody asking this question is impressed by which specific product you use; they're asking about the habit underneath it. Skip a rehearsed-sounding answer that avoids naming anything you'd actually check — an answer with no limits in it reads as either untrue or unconsidered, neither of which lands well.

Guardrails

• What's considered a normal or impressive answer to this question is shifting quickly and will likely look different again within a year or two — this is a way to structure an honest answer now, not a fixed script.
• Don't invent a task you don't actually do to have a better example ready — this question tends to get a natural follow-up, and an invented example usually doesn't survive one.
• The honest version of this answer will vary a lot by role and by how much autonomy you actually have over your own tools — adapt the three parts to what's actually true for you, not to what sounds most polished.

All 751 AI guides · JulieMango plans from £17/mo