The following is a summary of “Artificial Intelligence–Generated Scientific Literature: A Critical Appraisal,” published in the January 2024 issue of Allergy & Immunology by Zybaczynska, et al.
Review articles are pivotal in guiding medical decisions and pinpointing directions for future research. With the rise of artificial intelligence (AI), people are becoming more interested in how this technology could change how medical writing is put together. One tool from Open AI, called Generative Pre-trained Transformer (GPT-4) (Open AI Inc., San Francisco, CA), provides feasibility in writing medical literature quickly by following simple instructions.
It’s important to check the truth of the pieces made, especially in specialized areas like allergy and immunology. The GPT-4 Chatbot was told to write two 1,000-word reviews about hereditary angioedema and eosinophilic esophagitis. The authors critically reviewed these pieces using the Joanna Briggs Institute (JBI) tool for text and opinion. They also looked at areas of interest like the language, the quality of the references, and how accurate the content was. The language used in the AI-generated minireviews was well-thought-out and logically focused on the topic.
However, according to the reviewers of the AI-generated articles, the content was shallow, didn’t seem to result from an analytical process, was missing important information, and had the wrong information. Even though it was told to use scientific references, the AI robot mostly used free tools and made-up references. AI may change the way medical literature is put together. However, because AI tools are often used to create medical literature, especially on topics with few resources, they must be carefully tested and proven correct.
Source: sciencedirect.com/science/article/abs/pii/S2213219823011261