According to The Wall Street Journal, Google is already testing its Med-PaLM 2 AI conversation system at the Mayo Clinic and other institutions. It was introduced just a few months ago at Google I/O and is based on the PaLM 2 large language model (LLM) that powers Bard, Google's ChatGPT competitor.
In contrast to the original model, Med-PaLM-2 has been trained using questions and answers from medical license tests as well as a selection of demonstrations by medical experts. That makes it an expert in responding to health-related queries, and the report states that it is also capable of labor-intensive tasks like document summarization and data organization.
Google published a paper describing its work on Med-PaLM2 during I/O. On the plus side, it showed traits like "alignment with medical consensus," reasoning skills, and even the capacity to produce answers that respondents preferred over those produced by physicians. More unfavorably, it had the same accuracy issues that we have observed with other Chat AI models.
In collaboration with the healthcare software provider Epic, Microsoft is also creating medical AI chat technology based on OpenAI's ChatGPT. According to a March announcement from Google, these projects also include employing AI for cancer treatment and ultrasound detection. Both businesses assert that they do not train their models using patient data and have pledged to maintain the privacy of patient information. Microsoft voiced concern about doctors using its ChatGPT technology to enhance patient conversations last month.
Google stated in an internal email obtained by the WSJ that it thinks the modified model might "be of tremendous value in countries that have more limited access to doctors." Google has however acknowledged that the technology is still in its infancy. "I don't feel that this kind of technology is yet at a place where I would want it in my family's healthcare journey," said Greg Corrado, senior research director at Google. The technology, he continued, "takes the places in healthcare where AI can be beneficial and expands them by 10-fold."
Tags
News