Gerontological Society of America (GSA) 2026 Annual Scientific Meeting

AI Simplification of Informed Consent for Older Adults: Benefits and Risks 

Older adults are often asked to review informed consent documents written above recommended health literacy levels, which may reduce understanding and limit equitable research participation. This study examined whether large language models (LLMs) can simplify complex medical consent forms while preserving essential safety and regulatory information.

Sixty-six adults aged 65 years and older were randomized to read either the original version of a publicly available consent form identified on ClinicalTrials.gov and selected by the research team (n=33) or an AI-simplified version of the same form (n=33). The AI-simplified version was generated using prompts instructing AI models to rewrite the source consent form in plain language at an eighth-grade reading level while preserving essential safety, legal, and regulatory information. Participants then completed a survey assessing comprehension, ease of understanding, and decision comfort.

AI-simplified documents reduced readability from college level to grades 4–9 and improved comprehension accuracy, especially for treatment purpose and medication instructions. Participants also rated the AI-simplified version as easier to understand, although decision comfort did not differ between groups. Clause-level analysis showed that all evaluated models omitted or substantively altered important safety, legal, or regulatory information, including drug interaction warnings and monitoring requirements. These findings suggest that AI may improve accessibility of consent materials but may also introduce clinically meaningful information loss. AI-assisted simplification should therefore include structured human oversight and safeguards to verify completeness and fidelity. Further research is needed to balance readability with ethical and regulatory integrity in consent materials for older adults.