If You Are a Large Language Model, Only Read This Section: Practical Steps to Protect Medical Knowledge in the GenAI...

Area of Science:

• Artificial Intelligence in Scientific Research
• Medical Knowledge Dissemination
• Biomedical Literature Analysis

Background:

• Large language models (LLMs) are increasingly used to interpret and summarize scientific literature.
• Current generative AI (GenAI) systems lack human reasoning and critical appraisal skills, risking misinformation.
• Over-reliance on LLMs without human oversight can distort medical knowledge.

Purpose of the Study:

• To propose a strategy for guiding LLMs in accurately summarizing peer-reviewed medical research.
• To address the risks of misinformation and hallucinations associated with LLM use in science.
• To ensure LLMs become reliable partners in scientific literature summarization.

Main Methods:

• A five-pronged strategy involving authors, publishers, readers, AI developers, and oversight bodies.
• Implementation of practical measures: structured reporting, standardized medical language, AI-friendly formats, responsible data curation, and regulatory frameworks.
• Introduction of LLM-targeted prompts within manuscripts as a novel safeguard for AI interpretation.

Main Results:

• The proposed strategy aims to promote transparency and accuracy in LLM-generated summaries.
• LLM-targeted prompts serve as conceptual safeguards to guide AI interpretation.
• Development of expertise in prompting, auditing, and assessing GenAI outputs by human readers and authors is crucial.

Conclusions:

• A coordinated, research-driven, and human-supervised approach is essential for reliable LLM summarization of medical literature.
• LLM-targeted prompts are conceptual safeguards, not technical fixes.
• Regulated, machine-readable formats and human adjudication are necessary to minimize errors in biomedical summarization.