Google’s Med-PaLM 2, an AI software designed to reply questions on medical data, has been in testing on the Mayo Clinic analysis hospital, amongst others, since April, The Wall Street Journal reported this morning. Med-PaLM 2 is a variant of PaLM 2, which was introduced at Google I/O in Could this 12 months. PaLM 2 is the language mannequin underpinning Google’s Bard.
The paper additionally mentions research Google made public in May (pdf) exhibiting that Med-PaLM 2 nonetheless suffers from a few of the accuracy points we’re already used to seeing in giant language fashions. Within the examine, physicians discovered extra inaccuracies and irrelevant data in solutions supplied by Google’s Med-PaLM and Med-PalM 2 than these of different docs.
Nonetheless, in virtually each different metric, resembling exhibiting proof of reasoning, consensus-supported solutions, or exhibiting no signal of incorrect comprehension, Med-PaLM 2 carried out roughly in addition to the precise docs.
WSJ reviews clients testing Med-PaLM 2 will management their information, which might be encrypted, and Google received’t have entry to it.
In response to Google senior analysis director Greg Corrado, WSJ says, Med-PaLM 2 remains to be in its early levels. Corrado mentioned that whereas he wouldn’t need it to be part of his family’s “healthcare journey,” he believes Med-PaLM 2 “takes the locations in healthcare the place AI may be helpful and expands them by 10-fold.”
We’ve reached out to Google and Mayo Clinic for extra data.