In the interest of transparency and reproducibility, all materials necessary to replicate the findings reported across the four studies are made publicly available in this repository. This includes the synthetic datasets generated by each AI system, the analysis code used to produce all reported statistics, figures, and tables, as well as the prompting protocols and parameter configurations employed during data generation. /data gpt_plain.csv — Synthetic respondent-level data generated by ChatGPT-4 (GPT-4-Turbo) without demographic conditioning, serving as the primary dataset for Study 1. gpt_enriched.csv — Synthetic respondent-level data generated by ChatGPT-4 augmented with population-level distributions sourced from the Czech Statistical Office, used in Study 2 to assess whether external statistical grounding improves distributional accuracy. lakmoos_clones_500.csv — Synthetic respondent-level data generated by the Lakmoos AI neuro-symbolic platform (n = 500), constituting the primary datasets for Studies 3 and 4. Population-level reference distributions from the Czech Statistical Office and the Median research agency, used as benchmarks in Studies 1–3, are incorporated directly into analysis.py. Individual-level survey data collected via Ipsos Instant Research (Study 4) are not included in this repository, as their redistribution is restricted under the terms of the data collection agreement. /code analysis.py — Primary analysis script. Executes all statistical comparisons reported in the manuscript, including one-sample t-tests and chi-square goodness-of-fit tests, effect size calculations (Cohen's d, Cramér's V), and the generation of all tables and figures. gpt_generator.py — Data generation script used to produce the ChatGPT synthetic samples. Requires a valid OpenAI API key and replicates the generation procedure described in Studies 1 and 2. /prompts prompting_protocol.md — Complete documentation of the prompting procedures used in Studies 1 and 2, including the full prompt texts, model version (api-version: 2024-08-01-preview), temperature settings (0.7), and maximum token limits (1,800). /supplementary Supplementary tables and figures providing extended analytical results that complement the findings reported in the main manuscript.