(AI-powered recruitment with a tablespoon of reality)
That is what the hiring inbox feels like now. Candidates use AI to bake perfect-looking CVs. Employers attempt to use AI to sniff out which ones are real. Somewhere in the middle is a real person who just wants to do good work and a very human recruiter, trying to tell the difference between genius and generic.
From the candidate’s point of view, AI is a magic trick. You feed in your old CV, press a few buttons, and out comes a job-winning masterpiece. It is keyword optimised, typo-free, and, if edited correctly, humble enough to avoid suspicion. It’s the spell-check we always wanted, just smarter and faster.
But when everyone does this CV magic trick, then all applications start to look the same. Ten neat documents, all saying the right things, all shaped to mirror the job description as advertised. However, as recruiters, we can’t rely on perfectly baked CVs alone. We need signs of real skill with proof that a person built, fixed, or led something. i.e. who baked the cake versus who had ChatGPT do the shopping, baking, and icing.
On the employer side, their AI recruitment tools are scanning for patterns that smell robotic with the telltale signs of machine-made applications. Things like odd phrasing, overconfident buzzwords, % improvement stats, and even writing style mismatches compared to a candidate’s GitHub or LinkedIn posts. One study found that around 40% of applications for some tech roles were either fake or AI-generated. That’s a lot of digital cake mix! While these HR AI tools help, they also risk throwing out good people who didn’t match their CV directly to the job description or who missed a few key words. Detection is a filter, not a final verdict. It should narrow the pile, not make the decision. It is also important to remember that a selection process that screens out the right candidate is just as damaging as one that screens in the wrong candidate.
There’s a legal nudge here, too. Under the EU AI Act, hiring tools that rely on automation and AI are classified as “high-risk.” A human must stay in the loop to oversee the system, justify the outcomes, and be capable of explaining its decisions in plain language. That’s not just good practice, it’s the rule. So no, your AI-powered CV sorter cannot fire people on your behalf, at least not yet.
As with any major change or shift in technology, things get messy and many of us specialist recruiters are quietly enjoying a moment of clarity.
At CompuStaff.ie, we have seen this wave coming for a while. Our clients, many of them up to their eyeballs in beautifully written but utterly untestable CVs, are saying the same thing: “Just send us a few people you believe in.” Not thirty maybes but a small number of real-deal candidates we’ve spoken to, grilled (gently), and would personally vouch for. A shortlist that a hiring manager can review over a coffee, not a weekend. It’s a model of curation over chaos and evidence over adjectives.
Things are moving fast, and AI models are getting smarter every day, so what can we do now to utilise this technology to achieve the best outcomes?
My advice to candidates is to use AI for a first draft. Let it tidy up your CV, and then add the proof that only you can provide. Link to a GitHub repo, a project link, or a note about how you solved a messy backend issue. Include a new language you learned that helped you create a personal project. If AI helped you, say so. Honesty plus evidence beats flawless fiction.
If you are on the employer side, consider keeping the AI scanning tools, but pair them with a human test. A five-line code fix. A 10-minute screen share to walk through a decision a candidate made, or a simple system design sketch on a whiteboard. Aim for short challenges that are hard to fake and easy to judge. Do not ignore your “gut instinct” as AI cannot replicate your lived experience, intuition, or the subtle art of recruitment and selection that only comes through years of experience. Be sure to implement feedback loops and metrics into your recruitment system so your tools get smarter with each successful hire and not just faster. This will also ensure that you can explain its decisions, adapt when things go wrong, and prove it is fair and free from bias. Anything less is asking for trouble.
My hope for the next year is that the AI arms race in recruitment will level out. Smart candidates will get better at blending AI to help in creating their CV, and then add the links and evidence to prove their claims. Employers will stop expecting a perfect CV and look instead for signs of actual ability, curiosity, and craft.
Specialist niche recruiters like Compustaff will keep turning a sea of noise into a credible and curated shortlist that we will stand over for our clients.
So next time you’re staring at a batch of beautiful, beige applications, remember that some of those chocolate cakes were made with love. The others were made with a prompt.
Can you taste the difference?