About Bullshit Lingo, AI in HR, and Unmatched Needs.
A recruiter I don’t know personally posted on LinkedIn that she was done with bullshit lingo; the fluent, empty language that AI has poured into hiring on both sides of the table. Recruiters write vacancies with an assistant’s help. Candidates tailor CV’s with an assistant’s help. Somewhere in the middle sits a human who must see through both as to assess whether there’s a match, or if two AI-agents had had a conversation. Three of her team’s recent CV’s, she said, read like they’d been copied sentence for sentence from her own vacancy text. Her conclusion: AI accelerates everything except the judgment that actually matters.
My first reaction? Great, let’s make sure the human being behind the résumé comes to life.
I have opinions about vacancy texts. I’ve written about them before, about how often they miss the person entirely and settle for a wish list instead. So I wrote her a message; not a rebuttal, but a genuine question rooted in real life. I don’t fit a headline easily and my track record doesn’t sit inside one box. What would she advise someone like me, where a bit more nuance is essential but the HR system won’t capture it fully?
She answered generously and at length, and I believed every word was sincere. Pick your audience, and cut what isn’t relevant to the role in front of you. It takes discipline, she said, but it works. There’s one exception: if someone who already worked with you tells someone else, unprompted, that your differently-wired brain is exactly what’s needed, that route still gets you in.
I read her reply twice before I realised what bothered me. She’d just described the disease and prescribed more of it. If everyone narrows themselves to fit the shape of one vacancy, over and over, tailored fresh each time, it really isn’t the opposite of bullshit lingo. I’d say, it’s actually the production line of it. And the one exit she named for people like myself depends on a network I am not already part of. Which means the honest version of her answer wasn’t advice, really. It was a description of who currently gets to be seen as a person, and who has to first prove they’re a fit before anyone bothers finding out.
The Pattern and The Claim
I wanted to disprove my assumption that my first reaction might have stemmed from misinterpreting her bullshit lingo post. So, out of curiosity rather than suspicion, I read several posts on her public timeline. Each a personal observation and, at the core, emphasising the human side of her work. Each one was built the same way: a personal ritual, a cited industry trend, a named concept, and a question to trigger comments. And underneath nearly all of them, the same claim, worded in a half dozen different ways. Judgment is the one thing AI can’t touch. The narrative suggests she is, professionally and personally, its custodian.
I don’t think it’s a false perspective and I believe it’s sincere. But a claim built entirely around irreplaceable human nuance, delivered through a form so uniform it reads like a template, regardless of who (or what) actually typed it, is arguing against itself without meaning to. If judgment is the thing that resists being reduced to a formula, and the argument for that arrives pre-formatted into one, the medium has already undermined the message.
In Contrast
Which triggered me to look at the company website. Two vacancies from the same company, found the same week, made the pattern harder to wave away. The first asked, as a hard requirement, for deep experience in one specific system. Everything the role actually involved, described three paragraphs later, was built around a different system entirely. The “what you’ll do” section didn’t, at all, explain the things you’ll do. Instead, it was a list of finished deliverables in the grammar of a project report, stitched under a header that promised a description of daily work and delivered an inventory of outputs instead.
The second wasn’t really a vacancy, despite its header. No fixed location, no named client, but rather an open door into a “talent pool” for future assignments that may or may not exist yet. Underneath it, a checklist spanning every cloud platform, every deployment tool, and every flavor of the discipline it named. I guess that when a posting isn’t describing one real need, it has to cover all the possible ones instead. It strikes me as the opposite of the advice given to pick an audience and cut what’s irrelevant.
I’m sure nobody sat down and decided to mislead anyone. These vacancies simply followed the ordinary process of a system churning out what’s expected: documents assembled from other documents, using the standard HR lingo, the standard template, and then published without a final pass to check whether the content was actually coherent and all pieces put together still expressed what they needed to. It strikes me as something similar to what happened to the recruiter’s own three copied CV’s.
Résumé Research
None of this turns out to be one company’s habit. Digging a bit deeper, it turns out to be common practice. A study released last year tested twenty-four roles with identical candidate profiles, changing only whether the CV summary was written by a person or by an AI-agent. Screening models preferred the AI-written version four times out of five, regardless of what the candidate actually offered, and shortlisting rates moved by as much as sixty percent depending on which model was doing the reading. Meaning: tailoring your CV to match the vacancy isn’t what the sound advice would be anyway. Better advice would be to write it in a way so that the screening models prefer your résumé over others.
And then there’s something called a ghost job: a job posting where no intent of hiring actually exists. Roughly a quarter of workers in a recent UK survey believed they’d applied to one. These phantom vacancies cause confusion and lead to scepticism amongst applicants. At the opposite end from the bare checklist is a sibling problem: the vacancy that oversells instead of underdescribes, covering an ordinary role in inflated language and an inflated title that promises more than the pay or the work ever will. It’s a different failure, but with the same effect: the applicant has to dig through language built to manage an impression rather than describe a job.
The Mismatch Challenge
None of this is really about this one recruiter I happened to stumble upon, and it isn’t really about artificial intelligence either. It’s about a habit that predates both: a systemic approach to structuring a process where we no longer see the discrepancies between what’s being asked and what’s being expected. AI didn’t invent that habit. It just made the process cheaper to maintain, which makes the gap easier to hide and faster to spread, from a LinkedIn post to a job description to, and I’ll admit this plainly, a set of assessment frameworks I’ve been building.
In an attempt to catch these discrepancies, the frameworks I came up with, as it turned out on a closer look, had exactly the same kind of mismatch challenge hidden inside them. The talent assessment tool I built didn’t quite catch those, unless and until an operator actively recognises them and tells the tool the two halves don’t match. This is why the recruiter’s statement about judgment caught my eye in the first place.
It all boils down to the difference between recruitment and selection, between evaluation and matching, between knowing someone can do the job or guessing there’s a statistical chance of picking the right person.
An Honest Question
I don’t have a finished fix, and I’d be lying if I offered one. What I have is an idea I’ve been building toward for a while, and testing this week against exactly the two postings above: a process that has to reconcile a job description against itself before it gets published, that scores a requirement against the actual strategic need behind it instead of against a keyword, that asks for evidence instead of a checkbox, and that treats a gap in a candidate’s profile as something to invest in rather than a reason to flat-out dismiss them. And, more importantly, one that asks the question almost nobody currently seems to ask in either direction: not only whether the candidate is good enough for the company, but whether the company has earned the time and effort of a candidate.
That last part matters more than it sounds like it should. Every version of this conversation I’ve had, in a company’s job posting, in a recruiter’s careful advice, in my own tools before I checked them properly, ran in one direction only: what does the candidate need to become to be acceptable? Nobody was asking what the company would need to become to deserve them. In the meantime, we seem to be stuck in a model where the mould defines the fit, and the system defines the match.
The Dig
Human Resources once was, in Dutch, simply Personeelszaken (staff affairs), and somewhere in the translation to HR, a person became a resource; something you draw from because you pay for it, something that sits on a payroll and, in that sense, ‘belongs’ to someone else. I don’t think the people writing these vacancies think of candidates that way on purpose. I think the documents and the structures do it for them, quietly, efficiently, and ever more quickly.
I wouldn’t apply for either of those two job vacancies. Not because they’re asking for the skills I don’t have. And not because AI took the honesty away. Not because each follows the exact same template format. But because neither text ever once appealed enough on a human level to even consider whether I could tick all the boxes.
To put it in other words: a shovel gets handed over with an instruction to dig, and no indication anywhere of what’s being dug: a canal, a hole for a tree, or the place where you’ll eventually be asked to lie down indefinitely. The straightforward and transparent version of hiring would tell you what the dig is for, before you ever pick up that shovel.
https://www.linkedin.com/pulse/heres-shovel-dig-roland-biemans-eohse
Comments are closed