AI-Facilitated Evidence-Based Debate on the Risks and Benefits of E-Cigarettes
Year: 2026
Type: preprint
Abstract: This study looks at how generative artificial intelligence (GenAI) systems evaluate evidence-based arguments in debated health policy issues, using electronic cigarettes as a case study. Ten leading GenAI systems will independently score structured debate arguments that represent harm-reduction and precautionary views. These arguments will cover five key areas: smoking cessation and harm reduction for adults, youth initiation and gateway effects, comparative health harms, long-term health effects and uncertainty, and population-level public health impact. All arguments will be based on 16 peer-reviewed umbrella reviews of the e-cigarette literature. The study will evaluate cross-system reliability, position bias, and within-system consistency by using intraclass correlation coefficients, linear mixed-effects models, and test-retest correlations. The findings will give initial evidence on whether GenAI systems can be reliable and unbiased evaluators of contested scientific evidence. This proof-of-concept study has important implications for AI use in evidence synthesis, policy analysis, and decision support.
Institution: University of Maryland, College Park
Language: English
Subfield: Physiology
Field: Medicine
Domain: Health Sciences
Open Access status: gold
OpenAlex ID: W7201874506
Cites:
Cited by:
Related to: