"AI + Healthcare" New Model! It is reported that Microsoft collaborates with Harvard University, and Copilot accesses authoritative medical data

According to reports, a significant update for Microsoft's (MSFT.US) AI tool "Copilot" is set to be launched this month, marking the company's first collaboration with Harvard Medical School. The updated version of "Copilot" will source information from Harvard Health Publishing to answer questions related to healthcare. The report also states that Microsoft will pay a usage fee to Harvard.

Dominic King, Vice President of Health Business at Microsoft's AI division, declined to comment on the collaboration with Harvard University, but the company indicated that Copilot will provide answers that are closer to the information users receive from healthcare practitioners than any currently available information.

King stated, "It is crucial to ensure that people can access reliable, trustworthy health information that meets their language and literacy needs. Part of this work is to ensure we obtain this information from the right channels."

King pointed out that the goal is to help users make informed decisions regarding complex conditions such as diabetes. In the past, experts have warned against relying on chatbots for medical advice. A study conducted by researchers at Stanford University in 2024 showed that of the 382 medical questions posed to OpenAI's ChatGPT, the chatbot provided "inappropriate" answers to about 20% of them.

The data provided by Harvard Health Publishing includes content related to mental health, but the report noted that Microsoft has not disclosed how the updated "Copilot" system will handle such queries.

The interaction of AI chatbots with individuals suffering from mental illness has drawn close attention from lawmakers and medical experts. Reports indicate that ChatGPT has been implicated in some crisis events that led to patient hospitalization or death.

The report also noted that another feature in development will enable the "smart assistant" to help users find healthcare facilities near their residence based on their medical needs and insurance coverage.

Mustafa Suleyman, CEO of Microsoft's AI division, has listed healthcare as one of the key focuses of the division's work. He has also ramped up hiring at the internal AI lab, which competes with OpenAI. Reports indicate that in June of this year, Microsoft, which employs clinical staff, stated that the diagnostic accuracy of an AI tool it developed is four times that of a team of doctors, and its cost is only a fraction of that of a team of doctors.

Last month, Microsoft signed a preliminary agreement with OpenAI to redefine their relationship, paving the way for OpenAI to transition into a for-profit company.

However, the report noted that Microsoft still urgently needs to establish a certain degree of technical independence from OpenAI. Last week, Microsoft CEO Satya Nadella stated that he would delegate some responsibilities to a deputy in order to focus on the company's major AI projects The report adds that Microsoft established a dedicated department for consumer artificial intelligence and research in 2024, and is training its models with the goal of eventually replacing OpenAI's workloads. Achieving this goal may take several years. Microsoft previously stated that OpenAI "will continue to be our partner in cutting-edge models," and its philosophy is to use the best existing models.

In the field of consumer artificial intelligence, Microsoft lags behind OpenAI. According to data from research firm Sensor Tower, the downloads of the Copilot mobile application have reached 95 million, while ChatGPT's downloads have exceeded 1 billion. Copilot exists as a consumer-facing application and a virtual assistant in corporate enterprise tools. However, it currently relies heavily on OpenAI's models when responding to user queries.

In August, Microsoft announced that it had begun public testing of a self-developed artificial intelligence model that can be used for its Copilot chatbot system. Researchers and engineers (mainly those newly hired from Google's DeepMind artificial intelligence lab) are primarily focused on advancing Microsoft's own model development. The company has already used non-OpenAI models in its other software. The report also states that the company is now using models from OpenAI's competitor Anthropic to support AI tools in its 365 products.

Artificial intelligence has been a major source of revenue for Microsoft, largely thanks to its Azure cloud computing division. Both OpenAI and other companies use the services of this division to handle artificial intelligence computing tasks and training. Additionally, Microsoft provides artificial intelligence tools in its office and enterprise software products, which have millions of users