Using Generative Artificial Intelligence When Writing Letters of Recommendation

Artificial intelligence (AI) provides an opportunity to streamline tasks within academic medicine. Generative AI (genAI) models, specifically, have the capacity to generate new written content, follow detailed instructions for product improvement, and incorporate content from supplemental data sources. While a part of the professional responsibility of faculty in academic medicine, writing letters of recommendation (LORs) is often time consuming and repetitive candidate to candidate. Yet, crafting these letters well is paramount to convey an applicant's unique attributes in a time when pass/fail grading and remote interviews are increasingly common.In this article, the authors provide an approachable framework for the ethical use of genAI to assist with writing LORs in academic medicine. They briefly discuss the fundamental structure of genAI, the advantages between several genAI models specifically for the task of letter writing, privacy concerns that can develop when using genAI, iterative methods to develop effective prompts to craft letter drafts, personalization of finalized content, genAI use to identify bias, and appropriate documentation of AI usage.Once practiced, this process can prevent the need for shortcuts, such as copying and pasting from CVs or reusing previously written letters between candidates, that currently sacrifice letter quality to reduce writing time. Ethical use, privacy, and disclosure necessitate a deliberate framework for the use of genAI in letter writing. Future research is needed to inform the development of a specific AI model to generate LORs. The framework presented here provides faculty with the steps needed to begin incorporating genAI into their letter writing practice.