PaLM 2: Google’s AI chatbot that can answer medical questions

10th July 2023
Original Source:

At the I/O 2023 event, Sundar Pichai, CEO of Google, introduced PaLM 2, a large language model (LLM) that boasts improved multilingual, reasoning, and coding abilities.

Expanded Multilingual and Task Performance

During the announcement, Pichai emphasized that PaLM 2 is trained on 100 languages and excels in a wide range of tasks. The model is slated to be integrated into Workspace apps, as well as specialized versions like Med-PaLM for medical applications and Sec-PaLM for security purposes.

Testing at Mayo Clinic Research Hospital

According to a report by The Verge, Med-PaLM 2 has been undergoing testing since April at the Mayo Clinic, a non-profit research hospital in the United States. Google envisions that Med-PaLM 2 will be particularly beneficial in areas with limited access to doctors.

Early Stages and Expansion

Greg Corrado, Google’s senior research director, acknowledges that this tech is still in its early stages. While he wouldn’t personally rely on it for his family’s healthcare, Corrado believes that Med-PaLM 2 amplifies the potential of AI in healthcare by tenfold, targeting areas where it can provide valuable assistance.

Advantages of Med-PaLM 2 Over General Chatbots

Google claims that Med-PaLM 2 surpasses general chatbots like Bard, Bing, and ChatGPT in healthcare-related conversations. The model is trained on a curated collection of medical expert demonstrations, enabling it to deliver more accurate and specialized responses.

Data Control and Privacy Measures

As mentioned in the report, customers involved in testing this tech will have control over their data, which will be encrypted to ensure privacy. Google will not have access to this data. This approach aligns with recent updates to Google’s policy, which states that publicly available data will be utilized to train models like Bard, while maintaining transparency regarding user privacy.

Confirmation from Google

Google spokesperson Christa Muldoon confirmed the company’s use of publicly available information from the open web to train language models such as Google Translate. The updated policy reflects Google’s commitment to leveraging data responsibly and in accordance with privacy regulations.


Original Source:

× WhatsApp