Generative AI in Clinical Care

Large language models are trained on huge amounts of text to predict likely next words. With further tuning they can follow instructions, summarize, draft, translate and extract information. • They generate fluent text — but fluency isn't accuracy. • Grounding (giving the model the source documents to use) and retrieval-augmented generation (RAG) improve factuality. • Multimodal models handle images and audio as well.