Thursday, November 19, 2026← Weras Labs main siteTürkçe
WerasBlog

What's happening in AI, software, technology and science; plain and sourced.

Subscribe
AI

AI in Human Resources: Use Cases and Limits

HR teams can use AI to speed up everything from job postings to training materials. But in areas like candidate screening, bias and data protection risks demand real care.

AI in Human Resources: Use Cases and Limits
Image: Resume Genius · Unsplash · Unsplash License

AI in human resources is safest and most productive in text-heavy work: writing job postings and role descriptions, preparing structured interview questions, creating training and onboarding materials, and drafting employee communications. In decisions with real consequences for people, like candidate screening, performance reviews, or promotions, AI should stay in a supporting role and a human must make the call. The reasons are both the risk of bias and legal obligations under data protection law, which in Turkey means KVKK.

HR teams usually handle a heavy workload with a small staff: hiring, onboarding, training, payroll coordination, employee questions, policy documents. Since much of this work is writing and communication, AI can save a lot of time. But HR is also one of the most sensitive places to use AI, because every decision affects someone's career and life.

Low-risk areas for AI in human resources

Job postings and role descriptions

A good job posting clearly describes the real responsibilities of the role, the expectations, and what's on offer. AI is very effective at turning a hiring manager's scattered notes into a well-organized posting.

Write a job posting from the notes below.
Role: [title], [team], [work setup: office/hybrid/remote]
Key responsibilities: [notes]
Required skills: [notes]
Nice to have: [notes]
What we offer: [notes]
Rules:
- Don't use or imply discriminatory language about gender, age, or marital status.
- List "required" and "nice to have" skills separately.
- Avoid hype and clichés ("dynamic team", "rockstar", etc.).
- 350 words max.

The check for discriminatory language should be added on purpose. Phrases like "young and dynamic" carry both legal and reputational risk, and a model may produce them out of habit from its training data.

Structured interview questions

Structured interviews, where every candidate gets the same questions, make evaluation fairer and more comparable. AI can help you prepare behavioral questions based on the role's competencies, along with evaluation criteria for each question.

Prepare 8 behavioral interview questions for [role] that assess
these competencies: [list of competencies]. For each question:
- The competency it measures
- What we want to see in a strong answer
- Signs of a weak answer
Don't include questions about personal life, family, or health.

Onboarding and training materials

Starting from company policies, process documents, and FAQs, you can draft onboarding guides for new hires, short knowledge-check quizzes, or first-30-days plans. All you need to do is check that the information is current and consistent with company policy.

Employee communication

Policy change announcements, benefits updates, event invitations, and internal newsletter copy can be drafted quickly with AI. For sensitive announcements, such as restructuring or layoffs, the text must be finalized with leadership and, where needed, the legal team.

High-risk areas: candidate evaluation and performance

Why caution matters

AI models can reflect societal biases present in their training data. When evaluating a résumé, a model may be influenced by a name, school, address, or career gap in ways you never intended. That influence is often invisible: the model explains its reasoning in sensible-sounding sentences, while the real drivers go unnoticed.

There's also an important legal point. Under KVKK, people have the right to object to an adverse outcome that results from analysis carried out exclusively by automated systems, and the EU's GDPR contains a similar safeguard. Rejecting a candidate based solely on an AI score brings that right directly into play.

A safer approach

  • Keep the decision with a human: AI can summarize résumés or tabulate information against the posting's criteria, but an HR specialist or manager should make the screening decision.
  • Strip unnecessary information: Don't send the model personal details that aren't needed for the evaluation (photo, date of birth, marital status, address).
  • Write criteria in advance: Set evaluation criteria before the role opens and apply them the same way to every candidate.
  • Be transparent: Tell candidates how their data is processed, including the use of any AI-assisted tools.
  • Spot-check: If you use an AI-assisted pre-screen, regularly sample rejected candidates and have a person review them too.

The same approach applies to performance reviews. AI can help a manager organize notes kept over the year or phrase feedback more constructively, but the evaluation itself belongs to the manager.

Employee data and privacy

HR handles the most sensitive personal data in a company: medical reports, salary information, disciplinary records, family details. Much of this counts as special category data under KVKK. Practical rules:

  1. Don't upload documents containing health, disciplinary, or salary information to general-purpose AI tools.
  2. Remove names and personal details when drafting replies to employee questions.
  3. Check the data processing terms of the tools you use and the country where data is processed.
  4. Write a separate, clear AI usage guide for the HR team.

An internal assistant for employee questions

Many HR teams lose time answering the same employee questions about leave entitlements, benefits, or expense policy. An internal assistant grounded in company policies can handle some of these. To be safe, such a system should answer only from approved policy documents, send questions that depend on someone's personal situation (e.g. "how much leave do I have left?") straight to HR, and cite its sources in every answer. We build internal assistants like this as part of our custom software and AI integration services.

Employer branding and LinkedIn

The LinkedIn visibility of HR teams and managers plays an increasingly important role in attracting talent. Regularly sharing company culture, stories from the team, and open roles takes time. Tools like HookSlayer, which help people draft LinkedIn posts while keeping their own voice, can make this easier; still, every story shared must be true and written with the consent of the employee involved.

In employer branding content, AI helps most with brainstorming and editing drafts. Turning a project experience a team member described into a short post, or reworking a job posting into a more readable LinkedIn announcement, goes much faster. But content that speaks on behalf of employees should be based on their actual words and approved by them before it's published. Made-up employee stories do serious damage to an employer brand once they come to light.

Conclusion

AI in human resources saves HR teams significant time on text-heavy work, from job postings to onboarding materials. In decisions with consequences for people, like screening, performance, and promotions, the risks of bias and data protection violations mean AI should only assist, and the decision should stay with a human. Keeping sensitive employee data out of general-purpose tools and giving the team clear rules are the foundation of safe use.

#human resources#hiring#ai#data privacy#employee experience

From Weras Labs

We can help you with this

We turn what you just read into working software for your business. You talk to the person writing the code; price and timeline are clear from day one.

Frequently asked questions

Is it a good idea to screen candidates with AI?

It's not recommended. AI can be used to summarize résumés or organize information, but models can carry bias, and data protection laws like KVKK give people the right to object to adverse decisions resulting from automated analysis. A human should make the screening decision.

What should you watch for in AI-written job postings?

There should be no discriminatory language implying gender, age, or marital status, required and nice-to-have skills should be separated, and responsibilities should reflect reality. Writing these rules explicitly into the prompt helps.

Is it safe to enter employee information into an AI tool?

Sensitive data such as health, salary, and disciplinary information should not go into general-purpose tools. Removing names and personal details when drafting reduces the risk considerably.

Which task should an HR team start with when adopting AI?

Low-risk tasks that don't require personal data, like writing job postings, preparing structured interview questions, or creating onboarding materials, are good starting points.

About the author

Habip Dayan

Habip Dayan

Founder of Weras Labs

Habip Dayan is the founder of Weras Labs, a company that builds software products for professionals and teams. He has been building software since 2018 and writes about AI, cloud architecture and product design.

Comments

Be respectful; no ads or personal data. Comments that break the rules are removed.