A year ago, we wrote about how rigid keyword filters in applicant tracking systems can quietly reject good candidates. Since then, the conversation about recruiting technology has changed. Tools based on artificial intelligence, including large language models, are now being added to hiring software, and many companies are wondering whether they should use them.
The promise is real, but so are the risks. This article looks at both.
What AI screening promises
Unlike a simple keyword filter, AI-based tools can, in principle, understand meaning. They can recognise that “K8s” and “Kubernetes” refer to the same thing, or that experience building payment systems is relevant to a fintech role even if the job title is different. They can summarise long CVs, suggest questions for interviews and help organise large numbers of applications.
For teams that receive many applications, this sounds like a solution to the problems of traditional filtering. And in some respects it is. Better understanding of language can reduce the number of strong candidates lost simply because they used different words.
Pitfall one: bias at scale
AI systems learn from data, and hiring data reflects past decisions. If those decisions favoured certain profiles, backgrounds or career paths, a system trained on them may learn to do the same. The difference is that it will apply those preferences to every application, quickly and consistently.
This is not a theoretical concern. Bias in automated hiring tools has been discussed publicly for years, and it is one of the reasons regulators in Europe are paying close attention to the use of AI in employment decisions. A tool can appear neutral while repeating patterns that nobody intended.
Pitfall two: decisions nobody can explain
With a keyword filter, it is at least possible to see why a CV was rejected. With many AI tools, the reasoning is much less clear. A candidate may receive a lower score without anyone being able to say exactly why.
This creates problems on both sides. Recruiters cannot easily check whether the tool is working as intended. Candidates receive decisions without explanation. Under GDPR, people already have rights regarding automated decisions that significantly affect them, so companies need to think carefully about how and where these tools are used.
Pitfall three: confident but wrong
Language models are good at producing text that sounds convincing. A summary of a CV can read well and still miss an important detail or misrepresent someone’s experience. If recruiters rely on summaries instead of reading the source, errors can slip through unnoticed.
Pitfall four: the candidate experience
Engineers notice when a process feels automated. Generic messages, unexplained rejections and assessments with no human contact can make a company look distant. In a competitive market, good candidates may simply choose not to continue.
Using AI sensibly
None of this means AI has no place in hiring. Used carefully, it can save time on routine tasks and help recruiters focus on what matters. A few principles help:
- Use it to support people, not replace them. Let AI help organise, summarise and suggest, while people make the decisions.
- Check the results. Regularly compare the tool’s suggestions with human judgement, and look at who is being ranked low or excluded.
- Ask suppliers hard questions. How was the tool trained? How is bias tested? Can its outputs be explained?
- Be transparent with candidates. Tell them if and how automated tools are used in the process.
- Keep the human contact. A conversation with a real person remains the best way to understand a candidate.
The recruiter’s role
As tools become more capable, the role of the recruiter changes, but it does not disappear. Understanding what a team really needs, recognising potential in an unusual profile, and building trust with candidates are things that still depend on experience and judgement.
At Boomit, our Precision Hiring approach uses technology where it helps and keeps people responsible for every decision that matters. If you are thinking about how to bring new tools into your hiring without losing good candidates, we would be glad to talk. You can reach us at boomit.pt.