Google’s AI Is Taking Over Medical Research
Updated On: March 20, 2025
Google announced a suite of new products and features on Tuesday aimed at improving health outcomes through artificial intelligence (AI). The updates, unveiled at Google’s annual Check Up event, include enhanced health-related search capabilities, new medical records APIs, and open AI models designed to accelerate drug discovery. These innovations underscore Google’s commitment to leveraging AI to address critical healthcare challenges and improve accessibility to medical information.
The company is using advanced AI and ranking systems to improve “knowledge panels” for thousands of health-related topics. These panels, which already provide information on common ailments like the flu and colds, will now cover a broader range of conditions. Additionally, Google is adding support for healthcare queries in Spanish, Portuguese, and Japanese on mobile devices, making health information more accessible to non-English speakers.
Google’s new What People Suggest feature aims to provide users with insights from online forums and discussions, offering perspectives from individuals who have dealt with specific health conditions. For example, someone searching for arthritis-friendly exercises might see tips from others who have arthritis. While this could be helpful for finding relatable experiences, it also raises concerns about the spread of misinformation. The COVID-19 pandemic demonstrated how quickly unverified health advice—like anti-vaccine rhetoric or unproven treatments—can gain traction online, often with dangerous consequences. Similarly, this feature risks surfacing anecdotal advice that may not be scientifically sound or medically safe.
The challenge lies in ensuring the accuracy of user-generated content. Google’s AI, while advanced, may struggle to distinguish between credible advice and harmful misinformation, especially since misleading claims often gain popularity due to their emotional or sensational nature. To address this, Google must implement clear disclaimers and mechanisms to flag unreliable content.
Google has also introduced TxGemma, a collection of open AI models built on its Gemma framework. TxGemma is designed to help researchers predict the properties of potential new therapies, such as their safety and efficacy, by analyzing text and the structures of therapeutic entities like proteins and chemicals. The company plans to make TxGemma available to the research community later this month, encouraging collaboration and innovation in the field.
Additionally, Google unveiled an AI co-scientist tool built on its Gemini 2.0 model. This tool assists biomedical researchers in parsing vast amounts of scientific literature and generating novel hypotheses. For instance, researchers studying the spread of a disease-causing microbe can use natural language to specify their goals, and the AI co-scientist will propose testable hypotheses, summarize relevant literature, and suggest experimental approaches. While the tool is not intended to replace human expertise, it aims to accelerate research and uncover new ideas. Google is already collaborating with institutions like Imperial College London, Houston Methodist, and Stanford University to evaluate its potential.