If recent reports are to be believed, Google, the leading search giant, has been actively testing a medical chatbot since April to advance its AI capabilities. Dubbed Med-PaLM 2, this iteration is an adaptation of Google’s PaLM 2, a sophisticated large language model (LLM).
The trials of Med-PaLM 2 have taken place at Mayo Clinic located in Rochester, as well as at various undisclosed medical facilities.
According to an internal email obtained by WSJ, it is anticipated that the medical chatbot, specifically developed for addressing medical queries, will be of great advantage in regions where access to doctors is relatively restricted. The company predicts that Med-PaLM 2 will excel in managing discussions related to healthcare, surpassing other general-purpose chatbots like Google’s Bard, Microsoft’s Bing, and OpenAI’s ChatGPT.
Greg Corrado, the senior research director at Google, has emphasized that Med-PaLM 2 is currently in its initial development phases. He further expressed his reluctance regarding the involvement of this chatbot in his own family’s healthcare journey. Nonetheless, he believes that artificial intelligence can significantly enhance certain aspects of healthcare and asserts that the chatbot has the potential to expand these areas exponentially by ten times, as per the WSJ report.
As per Engadget, the PaLM 2 has showcased remarkable capabilities such as being in alignment with medical consensus, possessing the ability to reason, and even generating preferred answers over those provided by physicians. Remarkably, Google’s medical AI has already achieved an impressive 85 percent score on medical tests.
In April, the technology giant expressed its intention to collaborate with Google Cloud customers to explore the potential use of the medical chatbot for deriving insights from intricate and unstructured medical texts.
Moreover, Google hinted that Med-PaLM could assist with various other tasks, including crafting concise and detailed responses, summarizing documentation, and extracting insights from internal data sets and scientific knowledge.
The company published a study in May highlighting the limitations of Med-PaLM. The findings revealed that, compared to doctors, the medical chatbot exhibited some inaccuracies and irrelevant information. Conversely, Med-PaLM demonstrated a relatively equivalent performance to real doctors in presenting logical reasoning evidence and delivering consensus-supported responses. Notably, the AI chatbot displayed no indications of misinterpretation.
Via Cord Cutter News