large language models
The patent includes the company's safety-focused LLMs incorporated into its Polaris constellation architecture.
The large-scale study on open and closed models revealed LLMs exhibit noticeable variance when responding to different instantiations of the same question.
Med-Gemini-3D can read 3D scans, generate radiology reports and answer questions, while Med-Gemini-Polygenic can predict health outcomes based on genomic data.
The Nurse Advisory Council comes on the heels of the generative AI company's announcement in October that it launched its Physician Advisory Council.
In a study published in Nature, Google revealed that its generative AI technology answered medically related questions with 92.6% accuracy.
The company says its LLM has passed over 100 healthcare certifications and surpassed GPT-4 and other commercial models’ performance.