Google has embarked on a journey to refine and test its AI chat technology, Med-PaLM 2, at one of the largest healthcare providers in the United States – the Mayo Clinic. The Wall Street Journal revealed the collaboration is part of Google’s broader push to infuse AI solutions into the healthcare sector.
Med-PaLM 2: More than a Chatbot
Med-PaLM 2 is an offshoot of Google’s broader language model, PaLM 2, which powers Google Bard, the company’s direct competitor to OpenAI’s ChatGPT. Initially unveiled at Google’s I/O 2023 developer conference, the Med-PaLM 2 variant has been specially tailored for medical use. The chatbot is trained on questions and answers from medical licensing exams, enabling it to answer health-related queries accurately. Additionally, it boasts several skills beyond answering questions, such as summarizing medical documents and organizing research data. Google’s paper on Med-PaLM 2 reveals its alignment with medical consensus and reasoning capabilities, features that have received positive feedback from test users. In some instances, respondents even preferred the answers generated by AI over those produced by physicians. However, the AI technology is not without its pitfalls. It has exhibited accuracy problems, similar to other AI chat models, raising concerns over potential misinformation. The chatbot’s responses have sometimes included more inaccuracies compared to healthcare professionals’ responses.
Google vs. Microsoft: The AI Healthcare Race
Google is not the sole player pursuing AI solutions in healthcare. Microsoft is also developing medical AI chat tech, based on OpenAI’s ChatGPT, in partnership with healthcare software company Epic. Both tech giants have assured confidentiality, claiming not to train their models on patient data. The AI development race aims to support overburdened doctors and assist patients, especially in regions with limited access to medical professionals. However, both companies have faced hurdles along the way. In 2019, Google faced backlash for collecting detailed health records from millions of Americans without informing patients or doctors. Microsoft’s bioGPT AI, too, was reported to have presented wildly inaccurate content, raising concerns about the technology’s reliability.
The Mayo Clinic: A Fertile Ground for AI Experimentation
The Mayo Clinic, a non-profit healthcare provider treating over 1.3 million patients annually, began using Med-PaLM 2 in April. It has already been employing a Google-powered search system to examine vast amounts of patient data. The clinic’s history of medical breakthroughs and dedication to cutting-edge research makes it a perfect testing ground for Google’s AI technology.
Technology in its Infancy: An Ongoing Process
Google recognizes that its technology is still in the early stages. Google senior research director Greg Corrado noted that while the AI tech shows promise, he wouldn’t want it involved in his family’s healthcare journey yet. However, he believes that such technology could significantly expand the beneficial uses of AI in healthcare. In conclusion, Google’s Med-PaLM 2 represents an exciting leap forward in the intersection of AI and healthcare. While there are hurdles to overcome, the potential for such technology to assist healthcare professionals and improve patient outcomes is enormous. For more information about AI advancements in healthcare, check here.
