GAIDeT declaration generator
The generator writes a declaration of generative AI use for a journal article, conference paper, dissertation, course paper or thesis. It follows the international GAIDeT taxonomy (Generative AI Delegation Taxonomy): the author names the AI model and the tasks given to it. The Ministry of Digital Transformation of Ukraine recommends GAIDeT as the main form of disclosure.
-
Author & AI model
Choose the type of work, specify the corresponding author (or collective responsibility) and the exact AI model name and version.
-
Delegated tasks
Select delegated tasks from the GAIDeT taxonomy categories and include clarifying notes if required.
-
Standardized output
Generate the declaration and copy the standardized statement into your manuscript, as a section of its own or where the journal or department asks.
Work and responsible author
Choose the type of work and specify the author who delegated the tasks, or indicate collective responsibility of all authors
Choose the kind of work and give your details for the student's declaration
Generative AI model
Specify the exact name and version of the generative AI system used
Delegated tasks (GAIDeT taxonomy)
Select categories and specific micro-tasks delegated to generative AI tools under human supervision
Nothing found. Try another word.
1. Conceptualization
2. Literature Review
3. Methodology
4. Software Development and Automation
5. Data Management
6. Writing and Editing
7. Ethics Review
8. Supervision
Additional notes
Optional. Prompt specifics, tool versions or the bounds of human oversight; the Ministry advises naming how the output was checked
The university's declaration (point 2.1.3) asks why AI was used and for a clear line between your work and the AI's
Confirmations
Only the ticked items go into the text
Final declaration
Standardized text block ready to insert into your manuscript. For a Ukrainian journal choose UA
Paste the text into the work as a section, «Declaration of artificial intelligence use», or print and sign it
When and where to declare
Under the Ministry's recommendations (Annex 3) and Zaporizhzhia Polytechnic's declaration of principles on generative AI
- wrote, translated, paraphrased or summarised substantive text of the work;
- helped to frame research questions, hypotheses or arguments;
- generated or changed code, data, tables, figures or other material of the work;
- analysed, modelled or interpreted data behind the results and conclusions;
- a journal, publisher, conference or funder asks for it.
Spell checking and basic formatting that do not change the content usually need no declaration.
- As a section of its own, «Declaration of generative AI use», unless the journal names another place; its rules come first.
- If AI was part of the method, describe it in the Methods as well: system and version, period, settings, how the output was checked.
- In a thesis, the department's guidelines set the place.
- Where AI played a real part in the method, keep an internal record as well: the main checks, significant human changes, the grounds for decisions. Keeping or publishing every prompt is not required.
- fabricating or falsifying data, sources, images or results;
- paraphrasing other texts to hide plagiarism or self-plagiarism;
- doing assessed work with AI where its rules do not provide for it;
- passing personal or unpublished data to unapproved services;
- as a reviewer or expert, passing someone else's manuscript, proposal or review to an AI service unless the procedure allows it;
- listing AI as an author or co-author.
- every source AI suggested: that it exists and really supports the claim;
- figures, quotations and code, the way your field checks them;
- that you can explain the methods and results, including those where AI was used (for students, at the defence);
- that every co-author knows about the use of AI, agrees with the declaration and has approved the final version.
Using AI is not in itself a breach of integrity, and disclosing it is no sign that the author did less: what is assessed is the task, the human control and how reliable the result is. GAIDeT records what was given to AI, but does not decide whether that was acceptable, and does not replace the description in the Methods. An AI detector's percentage proves no breach and cannot be the only ground for a decision.
GAIDeT is the main form, not the only acceptable one: AID Framework, GAMER, NIST AI Use Taxonomy, ICMJE, WAME, COPE.
Documents
The university's rules on AI and the Ministry's recommendations
Academic attribution and sources
What GAIDeT is, how to cite the original paper, links
Zaporizhzhia Polytechnic National University
The university hosts the GAIDeT tools on its site for its staff, researchers and students. They are built on the open-source code of the taxonomy's authors (MIT licence).
GAIDeT (Generative AI Delegation Taxonomy) is a taxonomy for describing the tasks researchers give to generative AI. Instead of bans or AI-text detectors, it asks authors to declare openly what the AI did. It has two levels: 8 stages of research and the specific tasks within each; control and responsibility stay with the human.
