Myths series
AI alone decides who to hire — true or false?
Myths series · Article 7/10

The image of an algorithm sorting and eliminating candidates with no human involvement fuels distrust among candidates and recruiters — and it is exactly what European law seeks to prevent.
In practice, professional pre-screening solutions do not hire or reject anyone automatically: they collect structured answers, compare them to job criteria, and produce a summary for human decision. The legal line between decision support and automated decision is clear — and manageable.
If your current process has a human skim 200 CVs in one evening with no structure, you already have a decision-quality problem. Responsible AI adds structure and traceability — not autonomy.
What people say
« AI alone decides who to hire. »
In practice
Responsible hiring AI recommends and documents; recruiters or hiring managers always decide — with override capability, zero auto-rejection, and a full audit trail.
Decision support vs automated decision
The distinction is fundamental legally and ethically. Decision support presents structured information; automated decision produces an effect (rejection, final ranking) without meaningful human review.
AI pre-screening that generates a summary and proposed score is the first case — provided process and contractual configuration prohibit the second.
GDPR Article 22: three conditions
Article 22 applies when three elements combine: (1) a decision, (2) based solely on automated processing, (3) producing legal effects or similarly significant impact.
Exceptions (contractual necessity, Union or member-state law, explicit consent) are narrow in external hiring. The safe path: never satisfy the solely automated condition.
In practice: AI score is a recommendation; the recruiter clicks advance or do not advance after reading — action logged in the ATS.
AI Act Article 14: effective human oversight
Article 14 requires high-risk systems to enable effective human oversight — not symbolic. Supervisors must understand system capabilities and limits, detect anomalies, intervene, disregard or override output.
Require a readable summary interface (not score alone), access to source answer excerpts, and documented override capability from your vendor.
- Understanding: recruiter knows which criteria were assessed.
- Detection: alerts on incomplete or inconsistent answers.
- Intervention: manual change of candidate status.
- Stop: ability to disable AI on sensitive roles.
Step-by-step workflow
Step 1 — Candidate completes Lucy interview (15–20 min). Step 2 — Automatic transcription and criteria scoring. Step 3 — Summary available in recruiter dashboard. Step 4 — Recruiter reads summary within 48h; listens to audio excerpt if needed. Step 5 — Recruiter decides: shortlist, hold, or reject — with optional comment. Step 6 — Decision synced to ATS; candidate email sent (human or approved template).
At no step does the system send final rejection without explicit human action.
Never auto-reject
Contractually and technically ban auto-rejection below threshold. Even a low-scoring candidate may have contextual elements (career change, atypical path) visible only in the transcript.
Internal policy should state: No candidate is eliminated without human review of an interview summary or CV. That sentence reassures candidates and regulators.
Score overrides and audit trail
Every override must be possible and logged: who, when, prior score, new status, free-text reason. These logs are your defense in audits or complaints.
HiLucy retains summary, applied criteria, and associated human decision — reconstructing a candidate's full journey in minutes.
Key takeaway
AI informs; humans decide: GDPR Art. 22, AI Act Art. 14, and a no-auto-reject workflow deliver a lawful, trustworthy process.
Frequently asked questions
Can AI alone decide to hire or reject?
No. GDPR (Article 22) and the AI Act require human review for significant decisions. AI gathers and structures — recruiters decide.
Want to move from reading to action? See how Hi Lucy automates your voice AI interviews and your approach to interviews powered by artificial intelligence.