During Google’s I/O 2023 event, CEO Sundar Pichai unveiled the PaLM 2 large language model (LLM), which boasts improved multilingual, reasoning, and coding capabilities. The model, trained on 100 languages, offers a wide range of functionalities and is set to be integrated into Workspace apps, as well as specialized versions like Med-PaLM for medical purposes and Sec-PaLM for security. Recent reports indicate that Google has been testing an AI chatbot, powered by Med-PaLM 2, to provide expert answers to medical inquiries.
According to sources, the Mayo Clinic research hospital, along with other organizations, has been involved in testing Med-PaLM 2 since April. The goal is to utilize the LLM in regions with limited access to doctors. Greg Corrado, Google’s senior research director, noted that Med-PaLM 2 is still in its early stages and expressed caution about relying solely on AI for personal healthcare decisions. However, he believes that Med-PaLM 2 can significantly enhance healthcare practices where AI can be of great benefit.
Google claims that Med-PaLM 2 surpasses general chatbots like Bard, Bing, and ChatGPT when it comes to healthcare conversations. The model is specifically trained on a carefully curated set of medical expert demonstrations. It is important to note that customer data collected during Med-PaLM 2 testing will be under the control of the users, encrypted, and inaccessible to Google.
Recently, there has been discussion about Google’s data usage for training AI models, including Bard. Google confirmed that it uses publicly available information from the open web to train language models like Bard, aligning with its transparent privacy policy.
