Not long ago, a question about AI only came up if you were applying for a technical job. Now it can come up anywhere: a graduate scheme, a retail management role, a hospital administration post, an apprenticeship. Interviewers want to know whether you can work alongside tools that are changing how almost every job gets done.
The good news is that these questions are not a technical test. Nobody expects a marketing graduate to explain how a language model works. What they are listening for is the same thing they listen for in every other answer: good judgement, honesty and a willingness to learn.
This guide covers the AI questions you are most likely to meet, what each one is really asking, and example answers you can adapt to your own experience.
What interviewers are really listening for
Whatever the wording, almost every AI question is checking four things:
- You use it sensibly. You have tried AI tools for real tasks, and you can say what for, rather than talking about AI in general terms.
- You check the output. You know AI tools can be confidently wrong, and you have a habit of checking before you rely on them.
- You know where not to use it. You can name situations where you would not use AI, such as confidential information, personal data or work that needs your own judgement.
- You keep learning. The tools change month by month, and you have a way of keeping up that is not just reading headlines.
Keep those four in mind and most AI questions become much easier to answer well.
"How do you use AI at the moment?"
What they are asking: whether your use is real and sensible, or vague. This is the most common AI question, and the weakest answer is the most common one: "I use ChatGPT for everything."
How to answer: name one or two specific tasks, the tool you use, what it saves you, and how you check it. Two tasks done well beat a list of ten.
Example answer (graduate):
"Mostly two things. When I am reading a long report for my dissertation, I use an AI assistant to pull out the main arguments so I can decide which sections to read properly. And when I write something important, I ask it to point out where my argument is unclear. I do not let it write for me, because then it stops being my work, and I always check any facts it gives me against the original source, because it does get things wrong."
Avoid: claiming you use it more than you do. The follow-up question is usually "give me an example", and an invented example falls apart quickly.
"How could AI change the work you would do in this role?"
What they are asking: whether you have thought about the actual job, not just AI in general. This question rewards research.
How to answer: pick one or two real tasks from the job description, say how AI might help with each, and say what would still need a person. Then link it back to why you want the role.
Example answer (customer service team leader):
"Looking at the role, a lot of the team's time goes on answering the same questions by email. I could see AI drafting first replies to common queries, which would free the team up for the complicated cases. But a person should still check every reply before it goes out, and anything involving a complaint or a vulnerable customer needs a human from the start. The part of this job I am most interested in is coaching the team, and that does not change."
Avoid: predictions about whole industries disappearing. Stay with the tasks in front of you.
"Tell me about a time you used an AI tool to get something done"
What they are asking: this is a competency question, so answer it with the STAR method: the situation, your task, what you did and the result. The part most people leave out is how they checked the result.
Example answer (operations apprentice):
"In my placement, I was asked to tidy up a spreadsheet of supplier contacts with a lot of duplicates. Doing it by hand would have taken most of a day. I asked an AI tool to help me write a formula that flagged likely duplicates, then went through the flagged rows myself, because some suppliers had two genuine branches with similar names. It took about an hour, and my manager now uses the same approach every month."
Notice the check: the AI found the likely duplicates, and the candidate made the final decision. That is the detail interviewers remember.
"An AI tool gives you an answer you are not sure is right. What do you do?"
What they are asking: whether you trust AI blindly. There is only one wrong answer here, which is "I would use it anyway".
How to answer: say how you would check it, who you would ask, and what you would do if you still could not confirm it.
Example answer:
"I would treat it as a starting point, not an answer. I would check it against a source I trust, such as the original document or our own guidance. If it was something that mattered and I still was not sure, I would ask someone who knows, and I would say that I had used an AI tool, so they knew what they were checking."
"Tell me about a time you decided not to use AI"
What they are asking: judgement. Knowing when not to use a tool is as valuable as knowing how to use it.
Good examples to draw on:
- work involving personal data or confidential information, which should not go into a public AI tool
- a personal statement or application, where the point is that it is in your own words
- a decision about a person, such as feedback for a colleague, where a human needs to own it
- a task where checking the AI's output would take longer than doing it yourself
Example answer:
"When I volunteered at a charity, I was writing up notes from calls with people who had asked for help. It would have been quicker to paste the notes into an AI tool to summarise them, but they included people's personal circumstances, so I did not. I wrote the summaries myself. It took longer, but it was the right call."
"How do you keep up to date with AI when it moves so quickly?"
What they are asking: whether you are curious and have a habit, not whether you know the latest headline.
How to answer: describe one or two concrete habits, and something you learned recently that you actually used.
Example answer:
"I follow a couple of newsletters that cover new tools in plain English, and every few weeks I try one new thing on a real task rather than just reading about it. Recently I learned that I get much better results when I give the tool an example of what I want rather than a long description, and I now use that every time."
Avoid: "I read the news." It is true for everyone and tells the interviewer nothing.
"Your team wants to start using AI for a task you do by hand. How would you help?"
What they are asking: whether you would be a help or a blocker when things change, and whether you would think about the risks as well as the benefits.
How to answer: be positive, suggest trying it on a small part of the work first, compare the results with how it is done now, and agree how the output gets checked.
Example answer:
"I would want to try it on a small sample first and compare it with how we do it now, so we knew whether it was actually better. I would offer to write down the steps I currently follow, because that makes it easier to see what the tool needs to do and where a person still needs to check. And I would make sure we agreed who checks the output before it goes anywhere."
If you do not use AI much
That is fine, and it is far better to say so than to bluff. Be honest, show that you are open to it, and if you can, try it on one real task before the interview so you have something specific to talk about:
"Honestly, I have not used it much yet. I tried it last week to plan a revision timetable, and it was a useful starting point, though I changed most of it to fit my week. I can see it being useful in this role for first drafts, and it is something I would like to get better at."
An interviewer will take a curious beginner over someone who overclaims.
Mistakes to avoid
- Overclaiming. If you say you use it every day, expect to be asked exactly how.
- "It does everything for me." It suggests you are not checking anything, or not doing the thinking yourself.
- Forgetting about privacy. Never suggest you would put confidential or personal information into a public AI tool.
- Buzzwords without examples. "Leveraging AI to drive efficiencies" means nothing. One real task means a lot.
- Letting AI write your interview answers. Use it to prepare and practise, not to write a script. Memorised answers sound memorised.
Practise these questions out loud
Reading a good answer is not the same as giving one. These questions are now part of the AI Career Mentor interview, so they can come up when you practise a full interview, alongside the questions every interview asks. You speak or type your answer, it is scored honestly, and you see a stronger version to learn from.
If you have not tried it yet, you can answer one interview question on our homepage and see your score in under two minutes, with no sign-up.