AI Assistant Receives Expert Consultation from Nutrition and Health Experts, Ranking First in Qianwen Score

Madam Li is 68 years old, and she suffers from diabetes and weakened digestion. What to eat every day has become a challenge for the whole family. Her son downloaded an AI application for her. After entering "My mom has diabetes and a weak stomach, how should I plan her three meals a day?", the answer surprised her son— it not only listed dietary principles but also gave specific advice such as "replace white porridge with mixed grain porridge and egg custard for breakfast," "prioritize steaming, boiling, and stewing as cooking methods," and "chop vegetables finely and cook them soft."

This is the free answer provided by the Qwen APP.

More and more people are turning to AI when they feel unwell, but how reliable are these answers? Recently, the Center for Science Communication on Food and Health conducted a test: they selected three mainstream AI health apps in the market, simulated 10 scenarios related to nutrition, health, and food safety, used a new phone to generate content, and had five top experts conduct a blind test. The evaluation criteria were straightforward: whether the information was correct, whether it could be understood by ordinary people, and whether the advice could be practically followed.

The results showed that Qwen scored the highest, especially in terms of operability and understandability.

Recently, experts held an in-depth discussion on this test result and the future application of AI in daily health scenarios.

Chen Junshe, the chief consultant of the National Food Safety Risk Assessment Center, said after reviewing Qwen's answer: "It's not easy." His standard for judgment was practical: "Although AI cannot replace doctors in making diagnoses, in providing specific suggestions and telling you exactly how to do things, Qwen's answers are more detailed."

(Chief Consultant of the National Food Safety Risk Assessment Center, Chen Junshe)

Chen Wei, head of the Department of Clinical Nutrition at Peking Union Medical College Hospital, had a more direct observation: "The public has generally accepted using AI for health consultations." He has seen many patients who bring AI suggestions to confirm in his clinic. Good AI answers can save him a lot of time explaining basic knowledge, but bad ones can be "misleading."

(Head of the Department of Clinical Nutrition at Peking Union Medical College Hospital, Chen Wei)

The core strengths of Qwen are two: first, its underlying data comes from authoritative sources such as guidelines, textbooks, and literature, and through a graded knowledge base and data governance, it minimizes "AI hallucinations"; second, its answers aim to be "understandable, evidence-based, and actionable," rather than just throwing a pile of jargon at you, but instead telling you what you can have for breakfast tomorrow.

This test and expert discussions actually answer a question that ordinary people care most about: Can we trust the health advice given by AI?

The conclusion is: it depends on the scenario.

If you need to diagnose a disease, prescribe medicine, or make major medical decisions—go see a doctor, as AI clearly doesn't handle this. But if you want to know "how to adjust your diet if your blood sugar is high," "is this food rumor true," or "how to arrange three meals a day for an elderly person with weak digestion"—Qwen can already provide very good references.

(Director of the Center for Science Communication on Food and Health, Zhong Kai)

Zhong Kai, director of the Center for Science Communication on Food and Health, made a comparison: a good AI health assistant is like a nurse at a triage desk—knowing some medical knowledge, able to help you identify risks, clarify whether you need to go to the hospital, and which department to visit. For ordinary people who are confused by increasingly specialized clinical departments, this value is very practical.