Apple and OpenEvidence open a study on why people trust AI health answers
Apple is recruiting up to 200,000 US iPhone users to test a medical AI app, hoping to learn when people believe health chatbots and whether they should.
The short version
- Up to 200,000 US adults aged 21 and over can join the Apple Health AI Usage Study
- Participants use an app built with OpenEvidence, a clinical AI tool that doctors already use
- Sharing HealthKit data such as sleep, heart rate and lab results is optional
- Apple says study data will not be used to train AI models
- Surveys run monthly, and the project is planned for longer than a year
Why it matters
Millions of people already ask chatbots about their health, yet little is known about why they believe the answers. This study could shape how future tools show sources and handle uncertainty. Until results appear, treat any chatbot as a way to prepare for a doctor's visit, not a replacement for one.
What Apple announced
Apple said on Sunday that it is starting a research study on the trust people place in AI tools that answer health questions. As Cryptopolitan on MSN reported, the announcement came at the HLTH conference in Las Vegas, where Apple's vice president of health and fitness, Sumbul Desai, M.D., spoke about it.
The study looks at three things: what actions people take about their health after using such a tool, how well they understand the replies, and how good those replies actually are. Apple says the findings should help the wider industry build health AI more responsibly, not just Apple itself.
How the study works
Volunteers take part through an app made with OpenEvidence, a clinical AI tool already used by doctors. Its distinguishing feature is that each reply is tied to citations from peer-reviewed research, meaning studies checked by outside experts before publication. Sources named include top medical journals such as JAMA, Nature and The New England Journal of Medicine.
The project is planned to last longer than a year, with a survey every month. Participants can also choose to share data from Apple's HealthKit, the part of the iPhone that stores health records. That can include resting heart rate, lab results, sleep and activity.
Apple wants to know whether adding personal details makes answers more relevant and easier to follow. It also says that health tools built on large language models, the technology behind chatbots, often give an incomplete or unclear picture. The study will test how the information is presented, and whether the app helps people prepare for their own doctor visits.
Who can join and what happens to the data
Apple will accept up to 200,000 people in the US. Participants must be at least 21 years old and need an iPhone 15 Pro or newer that runs Apple Intelligence, along with iOS 26 or later.
On privacy, Apple makes several promises. Study data will not be used to train AI models. Questions are stripped of identifying details before being sent to a secure cloud provider run by a third party, and OpenEvidence deletes the questions and answers after 24 hours. Sharing HealthKit data and survey answers is optional.
The app flags certain worrying questions and points users toward a healthcare provider or other resources. Apple stresses that it does not diagnose or treat any condition and does not replace professional medical advice.
Why the timing matters
Robert M. Wachter, M.D., who chairs the Department of Medicine at UCSF, noted that patients and clinicians both use AI, yet the tools are developed and tested in separate worlds. In his view, linking them and studying carefully what works is the way forward.
The study arrives as consumer chatbots handle a huge amount of medical curiosity. OpenAI has said that people send ChatGPT more than 300 million health-related questions every week. In July, it opened ChatGPT Health to all US adults, one day after a Florida pastor sued the company. He claimed the chatbot told him not to see a doctor, advice he said nearly killed him.
Apple has been moving deeper into health as well. In September it released a Health Age score connected to a $119 blood test from Quest Diagnostics.
What to take from it
A confident, calm answer feels trustworthy, but sounding sure is not proof of being right. That gap between tone and accuracy is what a trust study can help measure. Tools that show their sources, and admit what they do not know, give you a way to check the answer yourself.
The reports we have do not say when results will be published, and a year-long study means patience is needed. In the meantime, the safest use of a health chatbot is to get ready for an appointment, for example by drafting questions to ask. Anything urgent needs a real clinician.
What to watch
- When Apple and OpenEvidence will publish results, since the reports give no date
- Whether the findings show that cited sources actually raise people's trust, or only make answers look more convincing
- Whether Apple widens the study beyond newer iPhones and adults 21 and over
Try it
Next time you ask a chatbot a health question, ask what sources the answer rests on, then check one of them. Use the reply to write questions for your doctor.
Sources and links
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by Cryptopolitan on MSN · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
