Google is already testing an AI-powered chatbot in a hospital

There are many professions that can benefit from the capabilities that chatbots based on generative AI technologies can provide on a daily basis, improving their efficiency.

In this regard, a new Wall Street Journal report is now indicating that Google is testing Med-PaLM 2, a specific version of its large language model PaLM 2, the same technology that powers Google Bard, its “ChatGPT”, in the well-hospital The well-known Mayo Clinic in Minnesota, United States.

Med-PaLM 2 test with a small group of patients

After a month of training and several weeks of testing, Google began running tests of its generative AI model on a small group of patients from the aforementioned hospital to continue improving the system, which is still in its initial stage.

It was in the middle of last May, just after the month of training, and coinciding with the celebration of Google I/O, when Google published Report on Med-PaLM 2 Pointing out the positive responses his chatbot version can give to the medical field in most cases through a series of aspects, so much so that a survey conducted with users preferred generative AI diagnostics to human ones.

There is still a lot to improve, in the initial stage

However, Google’s artificial intelligence was far from perfect, with a level of error that the company’s experts considered significant, resulting in inaccurate diagnoses or providing information in certain situations. AI models, although today this level of error is actually lower.

Med-PaLM-2 is not only trained on questions and answers from medical examinations and demonstrations by medical experts to provide only diagnoses of health problems, but also has a series of capabilities that will allow medical professionals to be more efficient in their work. Daily functions, being able, for example, to order reports and generate summaries in a very short time.

Bringing artificial intelligence to other medical diagnostic methods

Google is also looking at how it can combine AI with other technologies to make medical diagnoses, though it’s not alone, as Microsoft is also working to bring its AI technology into the health field.

Both companies have gone so far as to indicate that they train their models without using patient data, keeping their respective information confidential.

To help where doctors are lacking

For Google, according to the aforementioned post, its updated model would “could be of great value in countries that have limited access to physicians,” though it will still have to wait a while for the technology to be ready, because according to Greg Corrado , Senior Director of Research at Google:

I don’t feel like this kind of technology is yet in a place that I want in my family’s healthcare journey.

He adds that the technology:

It takes the places in healthcare where AI can be useful and expands them 10-fold.”

more information: Wall Street Journal