Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

AI news & research

News ·

Apple and OpenEvidence open a 200,000-person study of AI health answers

Apple has launched the Apple Health AI Usage Study in the US, with an app built on OpenEvidence's clinical AI. Up to 200,000 adults aged 21 and over with an iPhone 15 Pro or later can ask health questions, get answers with citations, choose to share HealthKit data, and rate the responses. Apple says study data will not be used to train AI models.

Millions of people already type their symptoms into chatbots, and nobody has good data on what they do with the answers. Apple wants to collect some. Its Apple Health AI Usage Study, launched in the US this week, will follow how people ask health questions, how much they trust the replies, whether they understand them, and what they do next.

The study runs through its own app, built on OpenEvidence, a clinical AI service used by doctors that draws on journals such as The New England Journal of Medicine, JAMA and Nature. Answers come in plain language with links to peer-reviewed sources. Up to 200,000 people aged 21 or over can enrol if they have an iPhone 15 Pro or later on iOS 26 or later. The app does not use Apple Intelligence or Siri; questions are typed, though the keyboard's dictation works. Participants can choose to share HealthKit data such as resting heart rate, activity, sleep or lab results, so the researchers can see whether personal context makes answers more relevant. The study will also test whether the answers help people prepare for visits to their own doctors.

On privacy, Apple says study data will not train AI models, and participants control what they share. The app includes guardrails meant to spot worrying questions and point people towards a clinician. Apple states plainly that the app does not diagnose or treat anything.

The scale is the interesting part. Most evidence on consumer health chatbots comes from small trials or from companies reporting usage, such as OpenAI's own figures on how many people ask ChatGPT about health. A study of up to 200,000 people, with ratings attached to each answer, could show where these tools confuse people or give them false confidence. It also gives Apple structured feedback before any health assistant of its own reaches the iPhone.

The open questions are about publication. Apple has said the study will inform design "for Apple and the industry", but not whether results will appear in a peer-reviewed journal, when, or with what data. Participants also skew towards people who own recent iPhones and volunteer for studies, which is not the whole population that asks chatbots about health.

Our take: Testing how ordinary people understand and act on AI health answers is overdue, and a study this size could produce real evidence. Apple has not said how or when results will be published, which decides whether this is science or product research.

Source: The Verge AI

More stories