AI in HR: CV Screening and the Ethical Rules
HR automation: the CV screening, the vacancy texts, the onboarding flows and AI's ethical boundaries; build a hiring system that is fair and fast.
One vacancy draws 150 CVs; give each 3 minutes and a full working day goes into mere reading. The result is familiar: the CVs get read diagonally, good candidates get sifted out by name-and-format accidents, and the rest of the process (the invitation messages, the onboarding documents) drowns in hand work. HR automation is that load's intelligent division; but with one comment: here the talk is of people's fates, and the ethical boundaries matter more than the technical possibilities.
This article builds four blocks: the CV screening (by the correct method), the vacancy-and-communication texts, the onboarding flows and the red lines.
Block 1: The CV screening; an assistant, not a judge
The correct structure is criteria-based: first you write the vacancy's 5–7 concrete criteria (the experience field, the tool skills, the language, the location), then the AI structures every CV against those criteria: "the criteria-fit table + a short summary + what stays unclear". The result is not a ranking but a map: 150 CVs turn into a table visible at one glance, and the decision is made by you. The structure to be banned is the "pick the best 10" task: the model can sift by hidden criteria (the name, the university name, the writing style), and that means invisible bias. The technical rules: minimise the personal data (the privacy laws apply to the candidate data too; the fitting-plan tool + the retention rule), let the borderline cases stay with a human (the "unclear" category is not an automatic reject but a hand review) and fairness to the candidate: the CVs weak in format and strong in content get saved precisely in this system; the AI gets forced to look at the criteria, not the text quality.
Block 2: The texts; from the vacancy to the rejection letter
| The text | The AI role | The human layer |
|---|---|---|
| The vacancy ad | The structure + the variants; attractive, clear language | The honesty of the real terms; the exaggeration ban |
| The candidate correspondence | The invitation, reminder, status templates | The tone check; the name-and-detail precision |
| The interview question set | The structured questions tied to the criteria | The field-specific depth |
| The rejection letters | The respectful, concrete template | The sending itself: a company that writes no rejections loses reputation |
This block's hidden gain is the candidate experience: the fast reply, the clear status, the respectful rejection; these are the details that make a small company look large as an employer brand, and with the hybrid rule they come nearly free.
Block 3: The onboarding; the first week's automation
The new employee's first week is a repeating process and an ideal automation candidate: the preparation checklist (the equipment, the accesses, the documents: automatic tasks to the responsible people; with the flow tools), the first-day pack (the welcome document, the team map, the first-week plan: automatic generation from the template) and the knowledge base access (an AI-searchable base for the new employee's questions: the "I was shy to ask" problem finds a technical solution). Its measurement is simple too: three questions to the new employee after 2 weeks: what was missing, what was excess, what confused; the answers renew the process.
Block 4: The red lines; the ethical frame
- The final decision always with a human: the hiring, rejection and dismissal decisions are not for AI to make; that is both an ethical principle and the regulatory direction forming in world practice.
- The transparency: if an automatic tool is used in the screening, not hiding it; the candidate's "talk to a human" road must stay open.
- The bias audit: the periodic check: is there a pattern in the sifted-out profiles (the age, the gender, the name type)? If a doubt exists, the criteria design gets fixed.
- Distance from the video-analysis hype: the tools "evaluating a candidate by facial expression" are scientifically dubious and ethically problematic; serious HR does not use them.
- The data lifetime: the unselected candidates' CVs get stored for a term and deleted; the "let everything stay in the archive" approach is a privacy breach.
Questions about HR automation
I'm a small company hiring 2–3 a year; do I need this too?
Not the full system — the selected tools yes: the AI help on the vacancy text, tabulating the incoming CVs and the onboarding checklist save time at low hiring volume too. The screening automation, though, makes sense with volume; 20 CVs get read by hand.
Candidates write their CVs with AI; doesn't the screening lose its meaning?
It changes, it does not lose it: the shiny text is no longer a signal (everyone's is shiny), which is why the criteria-based look (the concrete experience, the verifiable skill) grows even more important. The interview stage's weight rises: the practical task, the real-scenario questions; the work gets checked, not the text.
Should AI track the employee performance too?
A careful zone: the report-and-metric automation (the sales numbers, the panel view) is normal; the individual behaviour tracking (the message analysis, the activity counters) poisons the trust environment and carries legal risk. The rule: measure the result, do not surveil the person.
Which tool do I start from?
Not a special HR platform — the tools in your hand: the existing AI subscription (the CV tabulation, the texts) + a sheet (the candidate tracking) + the knowledge base (the onboarding). The specialised systems (the ATS) arrive at a yearly volume of dozens of hires; and the selection criterion there follows the same logic as the CRM choice.
Professional support
Want your HR processes automated correctly?
For diagnostics, priorities and implementation architecture, see the Business Process Automation service.
Sources and further reading
Where to verify the source
For the labour relations requirements:
- e-qanun.az: the Labour Code and the personal data rules
Continuing the topic
This line's neighbouring articles:
- The personal data rules
- The onboarding base
- The AI culture in the team
- The document flows
- Other articles on this topic
Make one change in your next hire: speed the ad's writing with AI and run the incoming CVs into the criteria table. You will see the time saving; and build the ethical frame from day one: fixing it later is many times harder.
I'm Anar Rustamli - a strategist, entrepreneur, and AI adoption leader working at the edge of growth, technology, and human thinking. Since 2016, my work has focused on helping businesses evolve in a rapidly changing digital landscape. I design growth systems, AI-powered workflows, and strategic frameworks that align performance with purpose. I believe real growth happens when strategy, data, and human insight work together - and my mission is to help businesses adopt AI in a way that strengthens both their results and their identity.
