Skip to content
Skjld Labs

AI in recruitment: the tasks get automated, but the meeting remains

Recruitment is at the front of a shift facing all of working life: AI agents are taking over tasks, not just text drafts. E24's podcast and recruitment expert Sverre Haugen point to where the line should be drawn, and why "AI sloppiness" has become a real competitive disadvantage.

3 min read Laget med AI-hjelp
AI in recruitment: two people with a speech bubble, while machines do the background work

Few industries feel the AI shift more directly than recruitment. The core of the craft is text, conversations and judgment, exactly what language models imitate best. That also makes the industry a good place to see where the line between machine and human should actually run, before the rest of working life faces the same choices.

In a recent episode of E24's podcast, recruitment expert Sverre Haugen of Meyer Haugen takes stock, and his analysis is useful far beyond the HR department. Here's what we noted, paraphrased, with our own assessments at the end.

From text suggestions to tasks

The first point is that the technology has passed a threshold. The previous generation of tools suggested text: job ads, emails, summaries. Today's agents perform tasks: they transcribe the interview, update the candidate log, map skills against the requirements profile and set up the next step. The shift from text level to task level is the same one now visible across knowledge work, and it changes the question from "can AI write this for me?" to "which parts of my job should happen in the background?".

The episode's most sympathetic example is also the simplest: when transcription happens automatically, the recruiter can put down the notepad and actually look the candidate in the eye. The automation didn't remove the human meeting. It gave it its space back.

"AI sloppiness" is the new disadvantage

The second point has been given an apt name: AI sloppiness. When anyone can produce plausible text in seconds, generic AI content, such as job ads with no sender, applications with no individuality, follow-ups with no warmth, becomes a marker you don't want. Recipients notice, and it costs trust.

The paradox is real: the cheaper mediocre text becomes, the more valuable text with clear ownership gets. The episode's advice is not to avoid the tools, but never to send anything you haven't taken ownership of yourself. That applies to job seekers, and at least as much to businesses.

Judgment still requires a human

The third point is where the line should be hardest: decisions about people. Models inherit biases from their data, and in recruitment, errors of judgment aren't just embarrassing. They can also be discriminatory and illegal. The episode's line is clear: AI can sort, compile and suggest, but the assessment of a human being must be owned by a human being, with critical thinking as the one skill nobody gets to delegate away.

Interestingly, the episode points out that as knowledge becomes commonplace through the models, value shifts to skills: the ability to apply, adapt and judge. For recruiters, that means more weight on mapping what people actually do, less on what they can recite. For everyone else, it means the CV increasingly competes with demonstrated skills.

Our rule of thumb

We recognize the whole picture from our own practice, and have arrived at a simple rule for where automation should stop: automate what happens around the meeting, never the meeting itself. Transcription, summaries, logging, preparation and follow-up in the background: by all means, fully digital. But the judgment, the decision and the actual conversation are not bottlenecks to be optimized away. They are the product.

It's the same lesson we've written about before in different clothing: AI amplifies your working habits, good and bad. Used as background machinery, it frees attention for people. Used as a substitute for judgment, it produces AI sloppiness with your signature on it.

Conclusion

Recruitment shows in concentrate what all of working life is facing: the tasks around the core work are being automated quickly, and that is both possible and sensible. But the core work, such as judging, choosing and meeting another human being, becomes more valuable, not less, when everything around it runs itself. Those who set things up so the machines take the background and the humans take the meeting get both done better.

Based on an episode of E24's podcast with Sverre Haugen of Meyer Haugen (June 2026). Our account is paraphrased. Listen to the episode for the full reasoning. The rule of thumb and the closing assessments are our own.

From article to practice

Want to put this to work?

Hands-on AI training. Courses and training in Norwegian, based on the tasks you have today.

Read about Hands-on AI training Book a free call
AI in practice

Read on.

Newsletter

AI in practice, one email a month.

What we've tested, one tip you can use, and what's coming next.

You'll get the Skjld Labs newsletter once a month, and can unsubscribe at any time. Privacy

First step

Thirty minutes. Free.

A no-obligation conversation about what you are facing. We will say so if AI is not the right answer. We are happy to meet you at Langes gate 8 in Sandefjord, or on video.