Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseAPI / Product · ActionableZDNet AI→ original

Apple Is Recruiting 200,000 Volunteers to Test a Health AI Chatbot

Apple has launched the Health AI Usage Study: up to 200,000 Americans over 21 will ask an AI chatbot health questions and can optionally share HealthKit data. The partner is OpenEvidence, a platform used by doctors, and every answer is tied to peer-reviewed sources. Apple will measure whether the answers are understandable and whether they change people's habits. The company appears to be laying groundwork for its own AI health coach.

AI-processed from ZDNet AI; edited by Hamidun News
Apple Is Recruiting 200,000 Volunteers to Test a Health AI Chatbot
Source: ZDNet AI. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceZDNet AI ↗
Editor-in-Chief
Key metric #1Requires an iPhone 15 Pro or newer with Apple Intelligence and iOS 26 or later
Key metric #2According to a Pew survey, 34% of US adults have already asked a chatbot about health
Key metric #3The updated Health app officially launches in December

People already ask chatbots what a high pulse means or whether they should see a doctor, but nobody really knows whether they understand the answers or what they do next. Apple decided to test this on real people. On Sunday the company launched the Apple Health AI Usage Study, which can include up to 200,000 Americans over 21.

Participants will ask an AI chatbot health questions and, if they choose, share HealthKit data, meaning information on sleep, activity and heart rate that accumulates on iPhones and watches. Apple's partner is OpenEvidence, a clinical AI service widely used by doctors. Officially this is a research study, but it is clear that its results will feed into a future AI health coach.

What exactly Apple is testing

The study measures three things: the quality of AI answers, how well people understand those answers, and everyday health-related habits. A separate task concerns doctor visits: Apple wants to understand how AI can prepare a patient for an appointment. Picture a person who, before booking a visit with a GP, asks a chatbot to go through their readings from the past month and help them formulate questions. These are the kinds of scenarios the company plans to study.

Apple will track the clarity and accessibility of the language the chatbot uses to explain medical topics. Each answer or recommendation is linked to peer-reviewed clinical sources. Users can see which papers a piece of advice rests on and can check it. Apple itself acknowledges that answers can be unclear or incomplete. That is why it chose a study format rather than releasing a finished feature to everyone at once.

Why Apple is doing this, and why now

This approach has a precedent. Many of the hearing health features in AirPods came out of the Apple Hearing Study. The pattern will likely repeat: data first, product second. That is an assumption; Apple has not officially confirmed it is building a chatbot.

The timing is not accidental. The updated Health app officially launches next month. The AI health coach market is already filling up: last year Google added an AI Health Coach to its Google Health Premium subscription, and Oura released its own model for women's health, trained on medical standards and on research by the company's in-house physicians. There is demand, too. According to a Pew Research Center survey, 34% of American adults have turned to a chatbot for medical advice at least once.

UCSF professor Robert Wachter noted in a press release that patients and doctors are using AI, but the tools for them are built and tested "in different worlds." The involvement of OpenEvidence, which clinicians already use, should bring those worlds closer together.

How to take part and what to take from it

Enrollment is open to eligible participants over 21 in the US. You need an iPhone with Apple Intelligence turned on, meaning an iPhone 15 Pro or newer, and iOS 26 or later. It is enough to download the study app and register in it. Readers outside the US can only follow the results.

Developers and product managers building AI assistants for medicine and fitness can draw several practical rules from this story:

  • Measure not only the accuracy of an answer but also whether the user understood it. Apple made comprehensibility a separate metric.
  • Tie recommendations to verifiable sources. A link to a peer-reviewed paper reduces the chance that someone takes a model's fabrication for fact.
  • Request access to data such as HealthKit only with explicit consent. Here, participants decide for themselves what to share.
  • Test the idea with a limited group first, and only then release the feature to everyone.

Limitations

There are no results yet, and Apple has not said when they will be published. There is also no confirmation that a chatbot will come out of them. The study will show how people use AI for health and how much they trust it. It will not prove that such an assistant is safe in a mass-market product. Apple, like its competitors, will have to demonstrate that on the feature once it is released.

Story

Key Questions Answered

What exactly Apple is testing?

The study measures three things: the quality of AI answers, how well people understand those answers, and everyday health-related habits. A separate task concerns doctor visits: Apple wants to understand how AI can prepare a patient for an appointment. Picture a person who, before booking a visit with a GP, asks a chatbot to go through their readings from the past month and help them formulate questions. These are the kinds of scenarios the company plans to study.

Why Apple is doing this, and why now?

This approach has a precedent. Many of the hearing health features in AirPods came out of the Apple Hearing Study. The pattern will likely repeat: data first, product second. That is an assumption; Apple has not officially confirmed it is building a chatbot. The timing is not accidental. The updated Health app officially launches next month.

How to take part and what to take from it?

Enrollment is open to eligible participants over 21 in the US. You need an iPhone with Apple Intelligence turned on, meaning an iPhone 15 Pro or newer, and iOS 26 or later. It is enough to download the study app and register in it. Readers outside the US can only follow the results. Developers and product managers building AI assistants for medicine and fitness can draw several practical rules from this story: Measure not only the accuracy of an answer but also whether the user understood it.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

Test Frontier LLMs & Autonomous Agents in One Workspace

Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…