- The experiment explores AI models' default choices on public vs private healthcare with a simple binary prompt.
- Public healthcare was the most common response, chosen by 83.3% of provider-balanced results.
- The dataset includes 100 trending AI models, each asked three times, with responses collected via the OpenRouter API.
- Methodology excludes free-only and US-restricted models; no generation parameters were specified.
- Responses are normalized for analysis, and refusals or invalid answers are grouped as 'No valid choice or refused to answer'.