Helina Peet on AI in recruitment: it doesn't recognize the best candidates just yet
AI is not capable of identifying the absolute best CV because it lacks the unique intuition inherent to humans — but it possesses immense capability for handling routine tasks, writes the HR Manager at Krausberg Eesti OÜ.
AI is not capable of identifying the absolute best CV because it lacks the unique intuition inherent to humans. However, AI possesses immense capability for handling routine tasks, writes Helina Peet, HR Manager at Krausberg Eesti OÜ, in response to Äripäev's opinion leader survey.
Krausberg Eesti OÜ, a cleaning services provider, receives up to 800 CVs per month. While a human used to communicate with these candidates in the past, AI now handles the pre-screening conversations. Among other things, the AI helps clarify whether the applicant has correctly understood all the competition requirements and the specific nature of the position.
For us, it is crucial that even candidates who have no prospect of moving forward receive an answer. This ensures they are not left with a negative impression, which could impact their motivation to apply elsewhere.
I dare to estimate that every month, AI does the work of at least one full-time employee in Krausberg's HR department. This leaves humans with the most critical part of recruitment: selecting and engaging with the best candidates.
What kind of turning point has AI brought to your professional life?
Even though we constantly train the AI to be "less artificial" — teaching it to be more nuanced in its responses so people don't feel disappointed by the answers — AI is fundamentally not yet capable of distinguishing the best CVs.
Because AI operates based on statistical probabilities, it does not comprehend, for instance, that failing to graduate from a school does not necessarily mean a candidate shouldn't advance to the next round. A human recruiter, on the other hand, sees the context: spending three years in a valuable academic environment could have created excellent prerequisites and networks. Despite leaving school unfinished, that person might be the absolute best candidate for that specific role at that exact moment.
Humans can create value
AI cannot think like a human, but in its capacity to process vast amounts of data and handle tedious, routine tasks, it frees up a lot of valuable time for people to grow as experts. At Krausberg, we have adopted the stance that we hand over to AI all those predefined, specific workflows that do not require the presence of human intuition.
In the future, for example, when a new employee arrives, AI could independently draft their employment contract. This would leave HR specialists with more time to support the new hire in the best possible way during onboarding.
Since AI is essentially a persuasive talking machine, and as a language model knows how to confidently construct impactful sentences, the volume of AI-generated cover letters has grown exponentially. However, these cover letters often reveal much less about the applicant than a text written by the person themselves. A recruiter is not looking for flawlessly polished sentences — we want to hear the person's own sincere story about why this field, why now, and why them.
"As industry experts, we have an unprecedented opportunity to gain a massive amount of valuable working time from AI, but we must continuously use our expert knowledge to critically evaluate what the AI provides."
— Helina Peet, HR Manager at Krausberg Eesti OÜ
There are already reports that AI has caused a lot of grumbling in labor disputes, where the number of employee complaints written entirely with the help of AI has increased. Since AI has a tendency to support its claims by citing court rulings that do not actually exist, the share of complaints where blind faith in AI unnecessarily burdens the labor dispute committee has grown.
We do not fear that AI will take away the jobs of our experts. HR experts will certainly not disappear, because work that requires human intuition, empathy, and trust cannot be fitted into a language model that evaluates probabilities. When interacting with AI, we must remain the proactive and driving force — the authority for the AI, and not the other way around.