Google Medical Robot and other dry doctors? Better than a doctor?
Robotics industry caprice.
Google's Med-PaLM2 AI model, which is said to have been tested in some hospitals since April, is a variant of the PaLM2 model, designed to answer questions about medical information. PaLM2 is the basic model of Google's artificial intelligence.
According to an internal email, Google believes the tool is particularly helpful in countries with poor medical infrastructure, such as where patient access to doctors is limited. Med-PaLM2, trained by medical experts, will be able to provide effective medical conversations better than general-purpose chatbots such as Bard ChatGPT. In a study released in May, Google said the Med-PaLM2 model still has some accuracy problems, a common problem with large language models, which sometimes provide answers that are more inaccurate and irrelevant than doctors. Nevertheless, on almost all other metrics such as inference, the model's performance is almost comparable to the performance of real-life doctors.
But Greg Corrado, Google's senior director of research, said Med-PaLM2 is still in its early stages. Corrado also noted that while it doesn't want AI robots to be part of its own home healthcare, it believes the model can be successful in healthcare.
Google's research says customer data used to test Med-PaLM2 will be encrypted and Google has no access to it. On July 1, Google also updated its privacy policy, stating that any publicly available resources or data will be used by Google to train its AI models without copyright dispute. In addition to Google, OpenAl is also actively deploying the medical business. Microsoft also teamed up with medical software company Epic in April to create tools that automatically edit patient information.
In addition, Microsoft's cloud service is testing OpenAl's GPT model, which enables it to summarize doctor's notes or generate reminders.
Robotics industry caprice.
