Take inspiration from AI innovation

Med-PaLM is a large language model ​that is designed to answer medical questions. ​It was built based on Google’s existing LLM PaLM, ​but it was adapted for answering medical questions ​through training with clinicians. ​The Med-PaLM model is exciting ​because it demonstrates the potential ​for a model that already exists ​like PaLM to be specialized ​through training with experts in the field.

A multimodal model is an AI model ​that can accept and learn from multiple types of input, ​such as images, video, or audio

•  ​AI tools that incorporate multimodal models ​can perform additional tasks beyond what a model ​that analyzes just one type of data can do
• Med-PaLM has a multimodal version called Med-PaLM M. ​As a multimodal model, Med-PaLM M can do more ​than answer text-based medical questions. ​It can interpret several types of biomedical data, ​including language, medical imaging, and genomic data.