In a move to enhance the credibility and accuracy of health information delivered through its Copilot AI assistant, Microsoft has partnered with Harvard Medical School. The collaboration aims to provide users with responses that more closely resemble guidance from medical professionals. This partnership also marks a step in Microsoft's effort to diversify its AI resources and reduce its reliance on OpenAI.
The updated version of Copilot, expected to roll out as early as this month, will leverage information from Harvard Health Publishing to address user queries on healthcare topics. As part of the agreement, Microsoft will pay Harvard a licensing fee for the use of its content. Harvard Health Publishing produces a wide range of health information, including research summaries and wellness advice, drawing from the extensive expertise of Harvard Medical School.
Dominic King, Vice President of Health at Microsoft AI, emphasized the importance of providing users with credible and trustworthy health information tailored to their language and literacy. While King declined to comment directly on the Harvard collaboration, he stated Microsoft's intention for Copilot to deliver answers that are more in line with information users would receive from a medical practitioner. "Making sure that people have access to credible, trustworthy health information that is tailored to their language and their literacy and all kinds of things is essential,” said King. “Part of that is making sure that we're sourcing that material from the right places".
The partnership with Harvard is expected to strengthen Copilot's credibility and improve the quality of its responses, particularly in sensitive healthcare areas. Copilot has primarily relied on OpenAI's models across Microsoft applications, including Word and Outlook. Microsoft has begun integrating AI models from other companies, such as Anthropic, and is developing its own AI models. This strategy aims to provide technological independence from OpenAI while still utilizing their advanced models.
Microsoft's commitment to improving Copilot's health AI capabilities extends beyond this partnership. Another feature in development would allow Copilot to assist users in finding healthcare providers in their vicinity, based on their healthcare needs and insurance coverage. Furthermore, Microsoft has been increasing hiring at an internal AI lab that competes with OpenAI, with healthcare as a priority. In June, Microsoft announced an AI tool capable of diagnosing diseases with greater accuracy and at a lower cost than a group of doctors.
The move to incorporate Harvard Health Publishing content into Copilot comes amid growing concerns about the reliability of AI chatbots in providing medical advice. A 2024 study by Stanford University researchers revealed that OpenAI's ChatGPT gave "inappropriate" answers to approximately 20% of 382 medical questions. The collaboration between Microsoft and Harvard aims to address these concerns by providing users with more accurate and trustworthy health information.
