Google has begun testing the AI model “Med-PaLM 2” in the medical system
Google recently disclosed its artificial intelligence model, “Med-PaLM 2,” conceived specifically for applications within the medical field. According to a report by The Wall Street Journal, this model has been undergoing testing in institutions such as the Mayo Clinic since April this year.
Med-PaLM 2 is a derivative of its predecessor, Med-PaLM, and is fundamentally based on Google’s large-scale natural language model, “PaLM 2,” launched this year. It is also the underlying model supporting the Bard service, therefore capable of comprehending user-spoken content, which allows it to rapidly perform corresponding operations.
As per internal correspondence, Google anticipates that Med-PaLM 2 will surpass its own Bard, along with competitors Microsoft Bing and OpenAI’s ChatGPT, in effectively comprehending information within medical systems.
However, given that current large-scale natural language models can still yield inaccuracies within AI-technology interactions, the application of Med-PaLM 2 and similar technologies within the medical field could potentially elicit skepticism and panic. After all, misjudgments may lead to diagnostic errors and potentially exacerbate a patient’s condition.
On the other hand, considering that medical practitioners can also make diagnostic errors, it appears the implementation of AI-driven medical decisions via Med-PaLM 2 could relatively reduce such occurrences. Yet, this claim may still require a more extensive period of validation.
The report further stipulates that all data content is encrypted during the medical system’s test usage of Med-PaLM 2, making it inaccessible even to Google.
At present, Med-PaLM 2 is still in its nascent stages of development. In accordance with Google’s previous assertions, there are no plans to supplant existing healthcare professionals. However, it is anticipated that this could significantly elevate the efficiency of medical services.
