Generative AI makes it easier for candidates to produce CVs and for companies to process applications. As a result, polished documents say less and less about a person’s real capabilities. Recruitment is shifting from judging presentation to verifying experience, judgement and the ability to use tools effectively. For Czech companies, this is not merely a technological change. It is also a reason to revise selection criteria, personal-data safeguards and accountability for decisions.
What AI is actually changing in recruitment
The label AI covers several different technologies in recruitment. Generative AI creates or edits content such as job advertisements, CVs, e-mails and interview notes. Automation performs predefined actions, for example scheduling interviews, sending replies or searching for keywords. Algorithmic assessment ranks, recommends or rejects candidates according to a model or score.
The closer a tool comes to making a decision about a person, the more important clear criteria, human oversight and the ability to review the result become. AI does not necessarily enter a company through an officially purchased HR system. Candidates use public tools to edit CVs, while recruiters use them to write advertisements or summarise interviews. The first step should therefore not be buying more software, but identifying where AI already handles candidate data and where it affects assessment.
Why traditional recruitment signals are becoming weaker
A CV, cover letter or rehearsed answer has never been direct evidence of job performance. Generative AI amplifies this weakness. In minutes, it can adapt a CV to the language of an advertisement, create several versions of a cover letter or prepare answers to common interview questions.
This can be described as inflation of recruitment signals: professional presentation is more accessible, so it carries less information. Using AI is not necessarily dishonest. A company should nevertheless avoid confusing the quality of a document with the quality of a candidate.
AI-text detectors are not the answer. Research points to their error rates and how easily they can be circumvented.[1] It is more useful to verify whether the candidate understands what they submitted, can make their claims specific and can apply the stated capability in practice.
How to verify capability rather than presentation
A company must first separate essential capabilities from requirements included merely out of habit. General phrases such as “independence” or “excellent communication” are not enough. The company should describe the situations in which the capability must be demonstrated and the result the person is expected to deliver.
In a structured interview, candidates for the same role receive the same core questions and their answers are assessed against a pre-defined scale. Questions should focus on specific situations: What problem did the candidate solve? What were they personally responsible for? How did they decide? What was the result, and what would they do differently today?
For selected roles, a short work sample is useful. The use of AI may be allowed where it reflects real working conditions. The follow-up discussion should establish what the candidate created independently and what they adopted, how they formulated the prompt, how they checked accuracy, which errors they found and why they chose the proposed solution. Any assignment must remain proportionate.
Research in personnel selection supports methods linked to job requirements, including structured interviews, work samples and professional tests. No method is infallible, however. Combining several relevant sources of evidence is preferable to overestimating the accuracy of a single score.[2]
What is specific to the Czech labour market
The Czech Republic has long ranked among the European Union countries with low unemployment. According to harmonised Eurostat statistics, the general unemployment rate in 2024 was approximately 2.6%, while the employment rate for people aged 20–64 exceeded 82%.[3][4]
The general unemployment rate is based on the Labour Force Survey and the methodology of the International Labour Organization. Registered unemployment reported by the Czech Labour Office covers people entered in its register. These are different indicators and should not be used interchangeably.
Low overall unemployment does not mean that every occupation and region faces a labour shortage. EURES points to recruitment difficulties in some technical, skilled-trade, construction, transport and healthcare roles, although conditions differ markedly by location and profession.[5] In contrast, some junior, administrative or content-oriented roles may attract plenty of applicants.
In the Czech environment, AI therefore does not have to serve only to reduce headcount. In some businesses, it can supplement scarce capacity and enable the same team to process more documents, orders or client requests.
According to the Czech Statistical Office, around 11% of enterprises with ten or more employees used at least one AI technology in 2024. The share exceeded two fifths among large enterprises and was substantially lower among small ones.[6] The statistic does not show how intensively companies use AI or whether it is used specifically in HR. It also misses some informal use of public tools by employees.
The International Labour Organization estimates that generative AI is more likely to transform individual tasks than eliminate entire occupations. Clerical and administrative activities based on processing text, data and documents are among the most exposed.[7] Exposure to technology is not a forecast that a job will disappear.
A recruiter may spend less time writing e-mails and more time defining criteria or verifying candidates. An accountant may move from routine processing towards checking and interpretation. The key capability will no longer be simply producing material, but also verifying it, placing it in context and taking responsibility for the result.
Legal essentials and a practical checklist
The EU Artificial Intelligence Act generally classifies systems intended for recruitment, filtering applications or evaluating candidates as high-risk. The same regime does not apply to every tool used in HR; a tool used only to improve the wording of an advertisement will not usually fall into that category.[7]
The Act entered into force on 1 August 2024 and its obligations apply in stages. Since 2 February 2025, organisations have been required to take measures to ensure a sufficient level of AI literacy. Where AI processes candidates’ personal data, the GDPR also applies. Article 22 provides specific protection against decisions based solely on automated processing where they produce legal or similarly significant effects.[8]
What every company should do
- map both official and informal use of AI in recruitment,
- define what data may be entered into approved tools,
- specify role requirements and consistent assessment criteria,
- use proportionate work samples and structured questions,
- explain the process and purpose of data processing clearly to candidates,
- review suppliers, data flows and the possibility of meaningful human intervention.
AI can accelerate recruitment, but responsibility for a fair and reasoned decision remains with the company.
Sources
- Weber-Wulff et al.: Testing of detection tools for AI-generated text — https://doi.org/10.1186/s41239-023-00424-3
- Sackett et al.: Revisiting meta-analytic estimates of validity in personnel selection — https://doi.org/10.1111/peps.12539
- Eurostat: Unemployment statistics — https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Unemployment_statistics
- Eurostat Data Browser: Employment rates — https://ec.europa.eu/eurostat/databrowser/view/lfsi_emp_a/default/table?lang=en
- Czech Statistical Office: Information Society in Figures 2025 — https://csu.gov.cz/produkty/informacni-spolecnost-v-cislech-2025
- International Labour Organization: Generative AI and Jobs — https://doi.org/10.54394/FHEM8239
- Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence — https://eur-lex.europa.eu/eli/reg/2024/1689/oj
- Regulation (EU) 2016/679 (GDPR) — https://eur-lex.europa.eu/eli/reg/2016/679/oj