Artificial Intelligence Detection Guidelines

IRMR screens all submissions for undisclosed AI content during initial review (3-5 days)

AI Policy Summary

PERMITTED: AI tools (ChatGPT, Claude, Grammarly AI) for language polishing, data analysis, minor drafting
PROHIBITED: AI as co-author or generating core research content without human oversight

Detection Process

  • Automated Screening: GPTZero, Originality.ai, Turnitin AI module
  • Red Flags: Uniform sentence structure, repetitive phrasing, fabricated citations, unnatural tone
  • Threshold: >20% AI probability triggers manual review

Thresholds and Actions

AI Probability Action
0-15% Auto-pass (cleared for peer review)
16-30% Manual review + AI disclosure request
31-50% Conditional return (7 days revise + disclose)
>50% Rejected; 3-month resubmission ban

Red Flags Triggering Rejection

  • Fabricated references/DOIs
  • Entire sections with identical AI "texture" (repetitive summaries, generic phrases)
  • Undisclosed heavy AI reliance (>30% content)
  • AI-generated images/figures without methodology disclosure

Outcomes

  • Pass: Advances to peer review with disclosed AI use noted
  • Conditional: Authors revise and document human contributions
  • Fail: Rejected with specific AI concerns; appeal with raw data/process logs
IRMR embraces AI as a tool but demands TRANSPARENCY. Disclose usage upfront for smooth review.