Wednesday, November 18, 2026

Top 5 This Week

Related Posts

Apple launches a giant study to find out if we can trust AI with our health


Millions of people already ask their health questions to ChatGPT, Gemini or Claude. But what do they really understand from the answers, and on what do they base their confidence? Apple wants to measure it on a large scale. THE November 15, 2026the group launched theApple Health AI Usage Studya study carried out with OpenEvidencea clinical AI platform used by healthcare professionals. Until 200,000 adults in the United States will be able to participate during more than a year.

What Apple wants to measure

The study does not seek to demonstrate that an AI gives good diagnoses. She is interested in user behavior. According to Apple, it must shed light on four questions:

  • what builds trust in the responses of a health AI;
  • the quality of these responses;
  • understanding what users actually have;
  • what they do next: change a habit, consult a doctor, do nothing.

Apple recognizes that consumer health tools based on large language models can provide answers incomplete or unclear. It is precisely to better present medical information that the company says it wants to do research, and it wants the results to guide the design of its own tools as well as those of the entire sector.

“Health is deeply personal, and conducting a dedicated research study is the most thoughtful way to understand how people use, understand and evaluate health AI on a daily basis,” explains Sumbul DesaiApple’s vice president of health and fitness.

Participants download a dedicated application and place their everyday health questions. The answers, written in simple language, are accompanied by clinical quotes verifiable. OpenEvidence draws in particular on the New England Journal of MedicineTHE Journal of the American Medical Association (JAMA) and Nature.

Participants can write down the answersgive their opinion and respond to questions monthly questionnaires. They can also, if they wish, share data from the iPhone Health application: resting heart rate, activity, sleep or analysis results. Apple wants to check whether this personal information makes the answers more relevant and easier to understand. The application must also help to prepare a consultation with his doctor. According to its profile on the App Store, it also allows you to place your personal health data in context and to build a personalized exercise or meal program.

The approach differs from that of general chatbots on one essential point: the application is not a commercial product, but a research tool. Each exchange becomes study data, analyzed to understand how a response is received, and not just whether it is accurate. This is what should allow Apple to measure, for example, whether the addition of personal data makes a response more relevant and more understandable.

Who can participate

Criteria Condition
Country United States only
Age 21 years and over
Device iPhone compatible with Apple Intelligence (iPhone 15 Pro or newer)
System iOS 26 or later
Number of participants Up to 200,000
Duration More than twelve months

The study is not not accessible in Francenor elsewhere in Europe, and Apple has not announced an extension.

What guarantees for the data?

Health is one of the most sensitive data. Apple highlights several commitments:

  • the data is anonymized before being sent to a secure third-party cloud provider;
  • they will not be used to train AI models;
  • OpenEvidence clears questions and answers after 24 hours;
  • participants decide for themselves whether or not to share their health data and their responses to questionnaires.

The application also includes guardrails: it identifies certain worrying questions and then invites the participant to consult a professional. Its App Store listing specifies that neither the application nor the study provide medical advice, diagnosis or treatment.

Why this is an important topic

Both patients and doctors are turning to AI, but with tools designed and tested separately. “Patients and clinicians are both turning to AI, but the tools they use are built and tested in separate worlds,” summarizes Robert Wachterwho heads the department of medicine at the University of California, San Francisco, cited in Apple’s announcement.

The risk is twofold. A person who places too much confidence in an incorrect answer can delay a necessary consultation. Conversely, a person who distrusts everything may miss useful information. Understanding what causes confidence to vary, the clarity of the response, the presence of sources or personalization, is therefore essential for designing safer tools.

For Apple, the issue is also strategic. TechCrunch notes that it is not yet established whether the group plans to integrate AI-generated health responses into its Health application or into Siri. The study could prepare this ground, on scientific bases rather than by a direct launch. More broadly, health questions are one of the most common uses of chatbots, as shown by the numerous articles in our health and medicine section.

Using a chatbot properly for your health questions

Without waiting for the results of the study, a few simple rules allow you to use AI assistants wisely:

  • use it to understand, not to decide: a chatbot can explain a medical term, an analysis result or the effects of a treatment, but does not replace either a diagnosis or a prescription;
  • favor sourced answers: an answer which cites studies or official recommendations is more easily verified than an assertion without reference;
  • prepare for your consultation: asking the AI ​​to list the questions to ask your doctor is a useful and risk-free use;
  • be wary of reassuring answers faced with worrying symptoms: in the event of chest pain, difficulty breathing or warning signs, you must call emergency services, in France on 15 or 112;
  • think before sharing your data: analysis results or history are sensitive information, which should only be entrusted to a service whose confidentiality policy is known.

Towards health AI evaluated in the field

With 200,000 potential participants and more than a year of follow-up, Apple’s study could become one of the largest ever conducted on the use of health AI by the general public. Its results, if published, will be of interest well beyond Apple: chatbot designers, health authorities and doctors are all seeking to know how to present medical information generated by AI clearly and without false confidence. One question remains for Europeans: knowing when, and under what conditions, such tools will arrive on their phones.

Shredder Smith
Shredder Smith
Shredder Smith is the lead curator and digital persona behind topaitools4you.com, an AI directory dedicated to "shredding" through industry hype to identify high-utility software for everyday users. Smith positions himself as a blunt, no-nonsense reviewer who vets thousands of emerging applications to filter out overpriced "wrappers" in favor of tools that offer genuine ROI and practical productivity. The site serves as a watchdog for the AI gold rush, providing categorized rankings and transparent reviews designed to help small businesses and creators navigate the crowded tech landscape without wasting money on low-value tools.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Popular Articles