Microsoft says people are using AI most often to ask about health-related topics

Microsoft has released a comprehensive analysis of how its Copilot AI tool is being utilized. Examining a sample of 37.5 million conversations from 2025, the research identifies health-related inquiries as the primary reason for AI usage. Users increasingly turn to AI via mobile devices for health advice due to its instant accessibility and ease of use. 

These findings align with the Digital 2026 report, which previously highlighted the intersection of AI, health, and internet trends. According to that report, the most popular activities for Generative AI (GenAI) and Large Language Models (LLM) involve seeking therapy and companionship. Additionally, generating solutions for a healthier lifestyle ranks among the top ten most frequent activities. 

The trend is equally prevalent in Turkey, where AI’s role in healthcare has become so significant that some news outlets have begun using the nickname “Dr. AI” to describe the phenomenon.  

Health is one of the top areas where people use AI in daily life. What about cancer patients?

The application of AI in oncology is both expanding and diversifying. It is utilized across the entire spectrum of healthcare, from clinical R&D and diagnostic to drug safety and patient adherence. 

However, research into the actual usage patterns of cancer patients remains scarce. While various AI tools are rigorously tested for accuracy and reliability, understanding how patients themselves engage with these technologies is a field still in its infancy. 

Cancer patients are cautious about AI 

According to a study of 666 cancer patients, although the vast majority (%98) are aware of AI, only %31 feel confident enough to explain the technology to others. Furthermore, %26 of participants reported using AI in the past, while only %5 specifically utilized it for health information or medical advice. 

In a study conducted with 508 cancer patients and their caregivers at the Ankara University Faculty of Medicine, Department of Oncology, frequently asked questions by cancer patients online were compiled and presented to artificial intelligence. The AI-generated responses were evaluated for quality and readability. Participants were selected from patients with the most prevalent cancer types (colorectal, breast, lung and prostate). 

The following are the most frequent questions asked online by patients, categorized by cancer type:

• “Where does breast cancer spread and metastasize?”
• “How is lung cancer staged and categorized, and what is its prognosis depending on the stage and type?”
• “What are the stages of prostate cancer, and how is it treated according to stage?”
• “What is adjuvant chemotherapy, and what are its benefits?”

The study indicates that AI may fail to provide satisfactory answers to specific queries regarding various treatment trials, primarily due to outdated datasets. Furthermore, the readability level of AI-generated responses was found to be higher (more complex) than expected for the average patient. Although AI is considered a strong candidate to bridge the information gap caused by the limited time health professionals can allocate to patients, the results suggest that it is currently too early for such a full-scale implementation.

Patient concerns regarding the use of artificial intelligence in healthcare

Patients prefer human physicians over AI during their treatment process. They seek direct communication with doctors regarding diagnosis, treatment continuity, and clinical inquiries, maintaining significant skepticism toward AI. According to a 2022 Yale University study, patients believe AI could be detrimental in terms of misdiagnosis, privacy breaches, reduced physician patient interaction, and increased healthcare costs. 

Similarly, a study conducted with 363 cancer patients at Ankara University Hospital in Turkey mirrored these anxieties. While patients believe AI can improve healthcare services to some extent (32%) or significantly (18.7%), they expressed concerns regarding misdiagnosis (31.1%), rising healthcare costs (27.5%), and the potential compromise of data privacy (19.6%). 

Limited Research on AI Usage Patterns Among Cancer Patients

An extensive body of literature has already emerged regarding AI’s role in medical imaging, interpretation, diagnosis, and the optimization of treatment processes. Furthermore, clinical trials led by healthcare professionals to evaluate the reliability and comprehensibility of AI-generated information are becoming increasingly common. Efforts to ensure that such information is accurate, high-quality, and accessible are progressing at an incredible pace. 

However, as with other healthcare services, these advancements inevitably lead us back to fundamental questions: access to care and implementation. Key areas that require deeper investigation include: Who in society can access AI? Which platforms and tools do they use, and with what frequency? Do patients trust or understand the information provided? Can they verify it, and how is it integrated into their treatment? Most importantly, how does this interaction affect treatment adherence and, ultimately, clinical outcomes? Future research must delve deeper into these patient behaviors and the practical application of AI.