1. AI in Recruiting - Just Not as Expected
It has long been clear that artificial intelligence would change recruiting. I mainly thought of the obvious applications: better search, faster research, support in preselecting and assessing candidates, or preparing documents. All of that is already happening, and I use tools like these in my work every day.
But now I have encountered AI in a place where I did not expect it: right in the middle of the job interview.
I first noticed it some time ago during a video interview. A candidate answered my questions in detail, in a highly structured way, and with remarkably polished language. Something about it bothered me. The answers sounded unusually formal and as if they were not being formulated in the moment; more like they were being read aloud. At first, I did not think much of it. But eventually I began to wonder: Is the person I am talking to reading answers that an AI is generating for them in the background?
2. When You Can See the Answer Reflected in Their Glasses
Since then, I have encountered this phenomenon several times. And apparently, I am not the only one. After interviewing one of my candidates, a client told me that she could actually see text reflected in his glasses, appearing on his screen in real time in response to her questions. What made this particularly surprising was that he was using AI support even for relatively general topics that a senior executive should normally be able to discuss confidently at any time, without much thought - and definitely without AI tools.
Over the past few weeks, I have noticed this more frequently, particularly during a current search in the US. In several interviews, individual answers sounded unusually polished and unnatural. So I began changing the way I conducted my interviews. I asked more spontaneous follow-up questions, shifted perspectives more quickly, and tried to ask questions that required an immediate response. Sometimes it worked. At other times, I still came away wondering: Was I actually talking to the candidate or, at least in part, to their AI?
3. The Test in the Face-to-Face Interview
With one candidate, things became particularly curious. Both my client and I had independently come away from the video interviews with the impression that some of her answers sounded as though she was reading them. But neither of us was entirely sure. And suspicion alone should not be enough to remove someone from a selection process.
Fortunately, the final step was an in-person meeting at the company’s office. The candidate arrived very early and had brought her laptop. As soon as she arrived, she asked for access to the Wi-Fi. The Office Manager gave her the password, assuming that she wanted to use the time before the interview to get some work done.
Then the interview began. The candidate positioned her small laptop rather discreetly in front of her and only glanced at it occasionally during the conversation. At first, the people sitting in the room with her noticed nothing unusual. One participant did, however. He had joined remotely, had already interviewed the same candidate before, and was therefore particularly alert to it. After he pointed it out, everyone realized what was happening: The candidate was using AI even during the face-to-face interview.
She was not confronted during the meeting, as no one wanted to embarrass her in front of the other participants. But she did not progress any further in the selection process.
4. What Are We Actually Assessing?
What interests me most about this story is not whether her behavior was right or wrong. I find the implications for the interview itself much more interesting.
Of course, candidates have always prepared for interviews. They have rehearsed answers, asked friends or coaches for advice, and taken notes. And there is obviously nothing wrong with using AI to prepare today. But there is a crucial difference between preparing with AI and having AI answer for you.
When I ask a candidate how they resolved a conflict, why they made a particular decision, or how they would respond to an unexpected situation, I want to understand how that person thinks, remembers, prioritizes, and makes connections on the spot. If AI takes over that process in real time, the interview itself becomes pointless.
5. Perhaps Interviews Need to Become Less Perfect Again
Perhaps good interviews will therefore need to emerge even more from the conversation itself. Ask follow-up questions. Take a thought in an unexpected direction. Ask about specific situations. Request details when answers remain general. Pick up on inconsistencies to find out how the person in front of us actually thinks, and whether they are even doing the thinking themselves.
Perhaps an unfinished sentence or a spontaneous answer will become more valuable again. Precisely because perfect answers are becoming easier and easier to generate.
6. My Impuls
How would you react if you realized during an interview that the person opposite you was having AI formulate their answers in real time?
Where, for you, is the line between legitimate support and distorting what a job interview is actually supposed to assess?
And do we need to change the way we conduct interviews when perfect answers tell us less and less about who actually formulated them?
We knew AI would change recruiting. What I did not expect was that we would no longer be able to be sure who we are actually talking to in a job interview.
At least not this soon.
