Policy on authorship and use of generative AI
- Artificial intelligence cannot act as an author or co-author, nor can it be cited as a source.
- If generative AI tools were used in the research (e.g., for data processing, modeling, statistical analysis, literature review, data management plans, translation, etc.), this should be noted in the “Methodology” section, with a description of the algorithms, datasets, and methods used.
- Data obtained using AI must be verified by scientifically sound methods.
- The use of AI should not replace critical thinking, human expert judgment, and adherence to ethical standards in scientific activity.
- The responsibility for the accuracy of information, avoidance of plagiarism, and reduction of the risk of bias associated with AI lies entirely with the authors.
- Concealing the use of AI is contrary to the principles of openness and ethics in science and may result in refusal to publish.
Methodology for checking texts for the presence/absence of fragments generated using AI
The editorial board uses the fiscal algorithms of the Undetectable and Copyleaks programs to check for AI Detector.
How to interpret the result and make sense of the Undetectable and Copyleaks infographics.
| Evaluation | Interpretation |
|---|---|
| (0-30)% Human | almost certainly the text was generated by AI |
| (40-60)% Human | mixed text (human + AI or edited AI text) |
| (70-100)% Human | probably written by a human (or carefully edited by a human) |
Features by which the programs Undetectable, Copyleaks, GPTZero and others determine the authorship of a text – whether it was written by a human or AI:
1. Complexity and variability.
The human writes unevenly (sometimes short, sometimes long), using synonyms, stylistic colouring, and sometimes making mistakes.
AI often writes fairly smoothly and competently, without jumps in complexity.
The algorithm measures this through:
• Perplexity – how unpredictable the text is to people.
• Burstiness – how much the length and structure of sentences vary.
The higher the perplexity and burstiness, the greater the percentage of human authorship.
2. Logical structure and style.
• AI uses clear paragraphs, consistent transitions (“finally”, “first”, “therefore”).
• A human’s structure is less formulaic, there are more individuality, emotions, comparisons, and reservations.
3. Phrase repetition and rhythm.
• AI tends to repeat the same words or constructions, especially synonyms.
• A human writes more diversely and intuitively, sometimes resorting to figurative and metaphorical means.
4. Semantic inconsistencies and “vivacity”.
• In texts written by humans, there are minor errors, unexpected examples, subjectivity and/or paradoxical thoughts.
• AI usually writes too smoothly “without roughness” – algorithms “see” this as signs of generation.
5. Statistical profile.
Services like Undetectable are trained on thousands of texts:
• which are marked as human and AI-generated;
• which compare your text with these samples on dozens of parameters.
Procedure for handling complaints regarding violations of academic integrity and ethics
The journal ensures a high level of quality in scientific publications and carefully considers all reports of possible violations of academic integrity and ethical standards — in particular, cases of fabrication or falsification of data, plagiarism, misattribution of authorship, etc. — in accordance with COPE recommendations. Responsibility for compliance with ethical principles and academic integrity rests entirely with the authors and co-authors.
Submitting a complaint
Complaints should be sent by email to the following addresses: kultdumka.iknamu@gmail.com (for authors – article submissions, publishing issues, etc.), kultdumka@icr.org.ua (for official correspondence – letters from organizations, etc.). The appeal must be clearly and specifically worded and contain confirmation of the violation.
Consideration of the complaint
Confirmation of receipt of the complaint will be sent within 10 business days, and the consideration process may take up to 60 business days.
If a violation is established, the following measures may be applied: refusal to publish the manuscript or withdrawal of an already published work; publication of an official statement describing the violation; notification of the author’s institution and relevant authorities; temporary ban on submitting materials to the journal.
If the violation is confirmed, the applicant and author(s) will receive a notification of the results.
All responsibility for compliance with academic integrity and ethics standards rests with the authors and co-authors.