T.R. ISTANBUL MEDENIYET UNIVERSITY
PRACTICE GUIDE FOR AI-SUPPORTED RESEARCH AND PUBLICATION PROCESSES
1. VISION: EPISTEMOLOGICAL TRANSFORMATION AND COGNITIVE PARTNERSHIP
Modern scientific research processes have reached a data volume that pushes the limits of human cognition in the age of "Information Explosion." Istanbul Medeniyet University defines Generative Artificial Intelligence (GenAI) not as a threat or substitute in this ecosystem, but as the researcher's "Cognitive Partner."
The fundamental working principle adopted in this partnership is the "Curation Model":
"Artificial Intelligence prepares the draft and rough structure (Drafter); The Researcher processes, verifies, gives scientific quality to the work, and assumes final responsibility (Curator)."
The aim of this guide is to free the researcher from mechanical and repetitive workloads, liberating the human mind for areas where it is truly competent: hypothesis generation, critical synthesis, and ethical reasoning.
2. ACADEMIC INTEGRITY AND LIMITS OF APPLICATION
At every stage of the research and publication process, the following "Permitted" and "Prohibited" limits determined in accordance with the IMU Academic Integrity Declaration are essential:
✅ PERMITTED: DRAFTING AND STRUCTURAL SUPPORT
(Manageable Assistance and Ideation)
Researchers may use GenAI tools within the scope of "Structural Support" for the following purposes:
- Ideation and Planning: Brainstorming, listing alternative hypotheses, and clarifying the problem statement.
- Technical Support: Receiving suggestions for language optimization, translation, literature review, code checking, data classification, and data cleaning.
- Analysis Support: Planning descriptive analysis steps, determining statistical method options, and creating visualization workflows.
- Drafting: Generating draft text blocks for specific sections of the study.
Conditions of Application:
- Scientific Responsibility: The final selection, justification, and scientific responsibility of the original research idea belong exclusively to the researcher.
- Verification: Analysis steps, assumption checks, and code outputs suggested by AI must be tested and verified by the researcher. Final analysis flow and interpretation responsibility lie with the author.
- Data Security: The "Data Minimization Principle" must be observed in sets transferred to AI tools for data analysis, and data must be anonymized to the extent possible.
- Recording: Tools used, version information, and basic prompts providing critical transformations must be recorded in study notes.
⛔ PROHIBITED: AUTONOMOUS GHOST-WRITING
(Production Without Human Supervision)
- Total Writing Ban: It is prohibited to use GenAI tools to write the entire text (from title to conclusion in one go).
- Originality Violation: Generated content cannot be used as final text in its raw form in the Hypothesis/Thesis, Method Design, Discussion, and Conclusion sections, which constitute the original contribution of the study. In these sections, AI can only be used for critical checking and presenting alternatives; final selection and comprehensive revision must be done by the researcher.
- Synthetic Data Ban: In accordance with CoHE ethical principles, it is strictly prohibited to generate synthetic data (fake survey responses, etc.) using AI instead of real participants in social research.
3. AI-SUPPORTED 6-PHASE RESEARCH PROTOCOL
PHASE 01: LITERATURE DISCOVERY
(Transition from Passive Reading to Active Verification)
Method: It is recommended to discover implicit connections among millions of articles with semantic search tools (Elicit, Scite, etc.) and use personal libraries (NotebookLM, etc.) as a queryable "Academic Memory."
⚠️ Critical Protocol (Triple Confirmation Chain): Due to the risk of hallucination (fabricating non-existent sources) in AI systems, every piece of summarized information must be verified in order: 1. AI Summary ➔ 2. Article Abstract ➔ 3. Full Text (PDF). No information whose original source cannot be accessed can be cited.
PHASE 02: HYPOTHESIS BUILDING
(Human-Centric Ideation)
Method: To optimize the research question, AI can be asked to act as "Devil's Advocate" to generate counter-arguments and identify logical gaps.
⚠️ Critical Protocol (Justification): The final selection and justification (Rationale) of the original research idea are done solely by the human researcher. AI cannot build a hypothesis, it can only test the established logic.
PHASE 03: DATA ANALYSIS
(Auditable Hybrid Intelligence)
Method: Data cleaning strategies, descriptive analysis plans, and visualization codes (Python/R) can be prepared with AI support.
⚠️ Critical Protocol:
- Anonymization: Protection of personal data (KVKK) is essential in data uploaded to AI tools. Data minimization must be applied.
- Testing: Codes and analysis steps written by AI must be verified by running them in the relevant software by the researcher personally. Unverified (Blind) analysis results cannot be reported.
PHASE 04: ACADEMIC WRITING
(Drafting and Curation)
Method: Draft text blocks can be generated for sections such as Introduction, Literature, or Method; language optimization and translation support can be obtained.
⚠️ Critical Protocol (Curation): AI outputs are "Raw Material." These texts cannot be used in the work without undergoing comprehensive editing (heavy editing), factual verification, and adaptation to the researcher's academic style.
PHASE 05: ETHICAL AUDIT
(Transparency and Documentation)
Method: Citation formats, structural integrity, and compliance with ethical boards can be audited with AI tools.
⚠️ Critical Protocol (Recording): In accordance with CoHE's "Openness Principle," tools used, version information, and critical prompts affecting the analysis result must be recorded in study notes.
PHASE 06: PUBLICATION STRATEGY
(Global Impact and Accuracy)
Method: Analyses for the most suitable journal (Scope Matching) for the article can be done with AI tools.
⚠️ Critical Protocol: The AI policies of the target journal must be examined. Maximum care must be taken to ensure that AI-induced biases do not affect the results during the publication process.
4. RESEARCHER CHECKLIST (SUBMISSION CHECKLIST)
Before submitting the study, it must be confirmed that the following criteria are met:
- [ ] Revision: Has significant editing and curation been done on the raw draft texts produced by AI?
- [ ] Verification: Has the accuracy of every piece of information, citation, and data in the text been personally verified by the researcher?
- [ ] Data Privacy: Has anonymization (Data Minimization Principle) been ensured in data uploaded to AI tools?
- [ ] Integrity: Is it guaranteed that absolutely no synthetic (fabricated) data was used in the study?
- [ ] Documentation: Have the tools, versions, and critical commands (prompts) used been recorded?
5. RESEARCH TOOLKIT AND INSTITUTIONAL SUPPORT
IMU AI Office guides researchers on the ethical use of the following tools:
| Phase |
Recommended Tool Categories |
Office Support |
| Literature |
NotebookLM, Elicit, Scite |
Hallucination Verification Training |
| Analysis |
Gemini Adv., ChatGPT (Data Analyst) |
Data Privacy and Prompt Workshop |
| Writing |
DeepL Write, Grammarly |
Academic Curation Seminar |
| Ethics |
Turnitin (AI Detection) |
Transparency Declaration Check |
© 2026 Istanbul Medeniyet University | Artificial Intelligence and Digital Transformation Office