Enhanced guidance on artificial intelligence for medical publication and communication professionals

The International Society for Medical Publication Professionals (ISMPP) position statement and call to action on the use of artificial intelligence (AI), published in 2024, recognized the value of AI while advocating for best practices to guide its use. In this commentary, we offer enhanced guidance on the call to action for ISMPP members and other medical communication professionals on the topics of education and training, implementation and use, and advocacy and community engagement. With AI rapidly revolutionizing scientific communication, members should stay up to date with advancements in the field by completing AI training courses, engaging with ISMPP AI education and training and other external training platforms, developing a practice of lifelong learning, and improving AI literacy. Members can successfully integrate and use AI by complying with organizational policies, ensuring fair access to AI models, complying with authorship guidance, properly disclosing the use of AI models or tools, respecting academic integrity and copyright restrictions, and understanding privacy protections. Members also need to be familiar with the systemic problem of bias with large language models, which can reinforce health inequities, as well as the limits of transparency and explainability with AI models, which can undermine source verification, bias detection, and even scientific integrity. AI models can produce hallucinations, results that are factually incorrect, irrelevant, or nonsensical, which is why all outputs from AI models should be reviewed and verified for accuracy by humans. With respect to advocacy and community engagement, members should advocate for the responsible use of AI, participate in developing AI policy and governance, work with underserved communities to get access to AI tools, and share findings for AI use cases or research results in peer-reviewed journals, conferences, and other professional platforms.