JetherVerse News
AI· 1 min read

Apple launches study on AI health tool accuracy

Apple and OpenEvidence are studying how well AI answers health questions, using a new iOS app and up to 200,000 US participants. Participants rate answers for accuracy, detail and relevance.

By JetherVerse News

Apple has launched a research study on how well AI tools answer health questions. The Verge reported the news on November 16, 2026. Apple runs the study with OpenEvidence through a new iOS app.

Up to 200,000 people in the United States aged 21 and over are eligible to join. Participants need an iPhone 15 Pro or newer running iOS 26 or later. The study does not use Apple Intelligence. Instead, participants ask health questions in the app and rate each answer for accuracy, detail and relevance.

Participants also choose whether to share data from Apple's HealthKit framework. Apple says the data will not be used to train AI models. The goal is to learn which answers people trust and which ones help them act on a health concern.

Sumbul Desai, Apple's vice president of Health and Fitness, said the company wants to understand how people use AI for health and how tools improve.

The study matters because people already ask chatbots about symptoms and medication. Ratings from a large group give developers real data on where answers fall short. Health apps built on language models face the same questions about accuracy, privacy and clear limits.

Few Nigerian users qualify for the study, since enrolment is limited to the United States. The findings still matter to anyone building health or wellness products for African markets. Phone-based triage tools are a growing category as clinic access remains uneven in many regions. Developers in Lagos, Nairobi and Accra building these tools face the same accuracy test Apple is now measuring at scale. Rating data from real users shows where a model misleads people before harm occurs. Participants rate three qualities: accuracy, detail and relevance. Builders should score their own products on the same three measures and track results over time.

If you build with AI models, test your answers against real user ratings before launch.

Source: The Verge

Get the news in your inbox

The tech stories that matter from JetherVerse News, with the source on every one.

Free. Unsubscribe any time.