The hype vs. the reality
Since 2023, the HR tech market has shipped dozens of tools that promise to "automate recruiting with AI." Most recruiters who have tried them report the same result: it works well for very specific tasks, and it produces considerable noise for everything else.
Where AI does move the needle
CV parsing. The most mature and useful application. A good parser extracts name, experience, skills and education in seconds. For an agency processing 50+ CVs a week, that is real time recovered.
Interview guide generation. Given a job description, current models generate reasonable-quality competency questions in 30 seconds. The recruiter adjusts in 2 minutes instead of building from scratch in 20.
Job description drafting. The first draft of a JD for a standard profile can come out of an LLM in seconds. Total time drops from 45 minutes to 10.
Where AI gets in the way
Automated candidate screening. Systems that generate a match score are optimising for keyword similarity, not real potential. A candidate with a poorly written CV but the right experience gets filtered out. The noise is high.
Automated video interviews. Platforms that analyse "body language" with AI have no solid scientific basis and generate immediate pushback from candidates.
Hiring decisions. AI can inform, not decide. Any system that proposes "hire" or "reject" without human review transfers risk without the level of reliability required.
The right question
Before adopting any AI tool: does it save time on a repetitive, low-judgement task, or is it trying to replace judgement on a high-impact decision? The first kind is worth it. The second kind creates problems.