Dataset: AI-generated references and RHS scores for chatbot bibliographic reliability analysis

This dataset contains AI-generated references and corresponding Reference Hallucination Scores (RHS) used in a cross-sectional study evaluating the bibliographic reliability of AI-based chatbots (ChatGPT, Gemini, and Perplexity). The dataset includes 3,150 references generated across multiple predefined subtopics and formats (letter and review), along with detailed scoring components (verifiability, bibliographic accuracy, PMID accuracy, and relevance). All prompts were originally entered in Turkish to reflect real-world user interaction. English translations are provided in the manuscript and supplementary materials. The data are provided in raw, unprocessed format to enable independent verification and reproducibility.