Google tests Med-PaLM 2 AI chatbot at Mayo Clinic

Google has embarked on an innovative initiative to enhance healthcare communication by testing its cutting-edge Med-PaLM 2 AI chatbot at the renowned Mayo Clinic and several other prominent hospitals, according to a report by The Wall Street Journal. Developed as an extension of the company’s PaLM 2 large language model (LLM), which powers the Bard, Med-PaLM 2 demonstrates significant potential in revolutionizing medical information retrieval and organization.

Unlike its predecessor, Med-PaLM 2 has undergone specialized training using medical licensing exam questions and answers, as well as a curated collection of medical expert demonstrations. As a result, it possesses a profound understanding of health-related inquiries and can efficiently undertake labor-intensive tasks such as document summarization and research data organization. These advanced features have the potential to streamline the delivery of accurate and relevant medical information.

During the Google I/O event, the company released a detailed research paper outlining the development of Med-PaLM 2. The paper showcased several commendable features, including its alignment with medical consensus, reasoning ability, and even the generation of answers preferred by respondents over those generated by physicians. However, it also acknowledged the lingering accuracy issues encountered in other Chat AI models, highlighting the need for further refinement.

Microsoft has joined forces with healthcare software company Epic to develop medical AI chat technology based on OpenAI’s ChatGPT. In a similar vein, Google is exploring the use of AI in ultrasound diagnosis and cancer therapy, as revealed in March. Both companies have pledged to maintain strict patient data confidentiality, emphasizing that their models are not trained on individual patient information. Recently, Microsoft expressed concerns about doctors utilizing ChatGPT to enhance patient communication, signaling the ethical considerations surrounding these technological advancements.

In an internal email obtained by the WSJ, Google expressed optimism about the potential value of the updated Med-PaLM 2 model in countries with limited access to healthcare professionals. However, the company acknowledges that the technology is still in its early stages and exercises caution regarding its integration into personal healthcare journeys. Greg Corrado, Google’s senior research director, emphasized the need for further development, stating, “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.” Nevertheless, he highlighted the potential of AI to augment healthcare services manifold in targeted areas.