Mapping the susceptibility of large language models to medical misinformation across clinical notes and social media: a cross-sectional benchmarking analysis
Key Points
This analysis aims to evaluate how susceptible large language models are to medical misinformation found in clinical notes and social media.
Cross-sectional study design
Evaluation of large language models
Comparison of responses to clinical notes and social media content
Benchmarking for assessment
Identified significant vulnerabilities in language models to misinformation
Differences in susceptibility found between clinical notes and social media content
Highlighted the need for improved training methods
Abstract
Scientific Computing and Data at Icahn School of Medicine and National Institutes of Health Office of Research Infrastructure.
Mapping the susceptibility of large language models to medical misinformation across clinical notes and social media: a cross-sectional benchmarking analysis | Synapse