Google's medical AI chatbot is reportedly being tested in hospitals

Med-PaLM 2 has been trained on a set of expert medical demonstrations.(Image: AFP)

Google's Med-PaLM 2 is reportedly being tested in various hospitals since April of this year.

The AI chatbot designed to answer medical questions is based on Google's PaLM 2, the Large Language Model (LLM) it announced at Google I/O in May. The model will also power Bard, its conversational and generative AI chatbot.

Also read | 5 Artificial Intelligence apps you can use for your sports training and fitness

As reported by The Wall Street Journal, which managed to get their hands on an internal mail, Med-PaLM 2 was trained on a set of expert medical demonstrations and has been designed to be used for Healthcare related problems.

In a blog post by Google in April of this year, the company said Med-PaLM 2, "was the first LLM to perform at an “expert” test-taker level performance on the MedQA dataset of US Medical Licensing Examination (USMLE)-style questions, reaching 85%+ accuracy".

It was also the first AI system to score 72.3% in Indian AIIMS and NEET medical examination questions.

Also read | Coming: The AI-led crackdown on money laundering

According to The Wall Street Journal report, Med-PaLM 2 will collect data from users testing it but it will be encrypted and Google will not have access to it.

The model also suffers from accuracy issues that plague other AI models but in tests, it performed more or less like an actual doctor.