Google is testing a medical AI chatbot at a leading US hospital

Google has begun testing Med-PaLM2 chatbot technology at one of the leading US hospitals, Mayo (Rochester, Minnesota). The technology is based on the PaLM 2 language model, which is superior to Google’s existing AI chatbot, Google Bard.

Med-PaLM2 is essentially a basic model that has undergone additional in-depth training on medical topics, based on questions/answers from medical licensing exams and medical examiner opinions. With this information, the chatbot can answer users’ questions about their health status, as well as summarize and systematize examination data.

In order to popularize the new product, Google has published a description of the process of its creation. There, some of Med-PaLM2’s features are revealed in detail, including its “ability to conform to medical common sense,” as well as its ability to reason and answer users’ questions. That said, in some cases, users found the AI chatbot’s answers to be more qualified than those of real doctors.

In an internal email, Google said the Med-PaLM2 technology is still quite “raw.” In the future, when the problems are fixed, its target audience will be users “in countries with limited access to health services.”

You may be interested: BMW has introduced ConnectedRide smart glasses for bikers