Adversarial prompting involves deliberately asking health-focused AI tools loaded or difficult questions to expose biases in how they treat different patients—whether they give worse advice for certain demographics, ignore important cultural factors, or reflect outdated medical assumptions. This direct testing approach helps you decide whether an AI tool is genuinely reliable for health decisions or whether it carries hidden risks.
Adversarial prompting involves deliberately testing AI tools with edge-case or identity-specific inputs to expose gaps, biased assumptions, or harmful outputs before relying on those tools for sensitive health decisions.
LGBTQ+ users, particularly transgender and intersex individuals, face elevated risk when AI health tools embed cisnormative or heteronormative assumptions, and adversarial testing helps identify which platforms are safe and affirming to use for medical research and care coordination.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.