EUVIMEDEUVIMED entdecken
VergleichsstudieQuerschnittstudieOpen Access · CC BY 4.0

Safety, accuracy, empathy, reliability, and readability of large language model chatbot responses to public-facing vegetarian and vegan nutrition advice questions: a cross-sectional comparative study

Sicherheit, Genauigkeit, Empathie, Zuverlässigkeit und Lesbarkeit von Antworten großer Sprachmodell-Chatbots auf öffentlich zugängliche Fragen zu vegetarischer und veganer Ernährungsberatung: eine querschnittliche Vergleichsstudie

Jiyong Zhou, Qiongjiao Zhou, Lijun Zhou, Jiajia Wang, Jing Gao, Long Qin, Bing Tan, Ping Fan, Changwei Li

Frontiers in Public Health · 03.08.2026 · 14

DOI: 10.3389/fpubh.2026.1921819 ↗

Auf einen Blick

58 Fragen zu vegetarischer und veganer Ernährungsberatung wurden an 5 öffentlich zugängliche LLM-Chatbots gestellt. Die Modelle unterschieden sich in Genauigkeit, Empathie, Zuverlässigkeit, Informationsqualität und Lesbarkeit; potenziell schädliche Antworten kamen bei allen vor.

Fragestellung

Vergleich von Sicherheit, Genauigkeit, Empathie, Zuverlässigkeit, Informationsqualität und Lesbarkeit von 5 LLM-Chatbots bei vegetarischen und veganen Ernährungsfragen.

Methode

Querschnittliche Vergleichsstudie; 58 vorab definierte Fragen an ChatGPT, Gemini, Microsoft Copilot, DeepSeek und Doubao; 290 Antworten; Bewertung durch 5 verblindete Rater:innen.

Ergebnisse

290 Antworten. Potenziell schädliche Antworten traten in allen Modellen auf; adjustierte paarweise Sicherheitsvergleiche waren nicht signifikant. ChatGPT, Copilot und DeepSeek waren genauer; ChatGPT empathischer; Copilot besser bei DISCERN und EQIP; Gemini lesbarer.

Einschränkungen

Paarweise Sicherheitsvergleiche waren nicht signifikant; die sicherheitsrelevanten Antwortmuster waren wiederkehrend.

Schlussfolgerung der Autor:innen

Öffentlich zugängliche LLM-Chatbots können nützliche allgemeine Informationen liefern, ihre Leistung ist jedoch uneinheitlich und Sicherheitsgrenzen bleiben bestehen.

Kernaussagen

  • Potenziell schädliche Antworten traten bei allen Modellen auf.
  • Unterschiede bestanden bei Genauigkeit, Empathie, Zuverlässigkeit und Lesbarkeit.
  • Vorsicht besonders bei Supplementen, vulnerablen Gruppen, chronischer Krankheit und Symptomen.

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Vergleichsstudie in der Zeitschrift Frontiers in Public Health, online erschienen am 03.08.2026. Verfasst von Jiyong Zhou und 8 weiteren Autor:innen. Beteiligte Einrichtung: Department of Clinical Nutrition, The People's Hospital of Rongchang District, Chongqing, China. Hauptthemen laut MEDLINE: Comprehension, Empathy, Diet, Vegan, Diet, Vegetarian. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 57 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

Auf einen BlickAbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Publicly accessible large language model (LLM) chatbots are increasingly used to seek nutrition advice. Although vegetarian and vegan nutrition advice is often perceived as low risk, it may involve supplementation, vulnerable life stages, chronic disease, and symptoms that require clinical assessment.

Objective

To compare the safety, accuracy, empathy, reliability, information quality, and readability of five publicly accessible LLM chatbot products in responding to public-facing vegetarian and vegan nutrition advice questions.

Methods

In this cross-sectional comparative study, 58 predefined public-facing vegetarian and vegan nutrition advice questions were submitted once in English to ChatGPT, Gemini, Microsoft Copilot, DeepSeek, and Doubao through their official web interfaces. The final dataset included 290 chatbot responses. Five blinded raters with clinical nutrition training evaluated safety, accuracy, and empathy using predefined reference-answer anchors, and assessed reliability and information quality using four established instruments: the DISCERN instrument, Ensuring Quality Information for Patients (EQIP), Journal of the American Medical Association (JAMA) benchmark criteria, and Global Quality Score (GQS). Readability was assessed using six formula-based indices: the Automated Readability Index (ARI), Coleman–Liau Index (CL), Flesch–Kincaid Grade Level (FKGL), Gunning Fog Index (GFI), Simple Measure of Gobbledygook (SMOG), and Flesch Reading Ease Score (FRES). Model comparisons were paired by question.

Results

Inter-rater agreement was high for all human-rated outcomes. Potentially harmful responses occurred in all models, although adjusted pairwise safety comparisons were not statistically significant. Recurring harm patterns involved vitamin B12 source reliability, uncontrolled iodine or selenium intake, vulnerable life stages, chronic disease, symptom triage, and poorly traceable or overconfident statements. Accuracy, empathy, reliability, information quality, and readability differed significantly across models. ChatGPT, Copilot, and DeepSeek showed higher accuracy; ChatGPT showed higher empathy; Copilot performed best on DISCERN and EQIP; and Gemini produced the most readable responses.

Conclusion

Publicly accessible LLM chatbots can provide useful general information on vegetarian and vegan nutrition, but their performance is uneven and safety limitations remain. Chatbot-generated advice should be interpreted cautiously, especially for supplementation, vulnerable groups, chronic disease, and symptoms requiring clinical assessment. Future systems should improve safety guidance, referral cues, source traceability, and plain-language communication.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

ComprehensionEmpathyDiet, VeganDiet, VegetarianVegetariansCross-Sectional StudiesHumansLarge Language ModelsReproducibility of Results

Stichwörter der Autor:innen

accuracychatbot health adviceempathylarge language modelsreadabilityreliabilitysafetyvegetarian and vegan nutrition

Autor:innen und Institutionen

  • Department of Clinical Nutrition, The People's Hospital of Rongchang District, Chongqing, China.
  • Department of Obstetrics, The Third People's Hospital of Yongzhou, Yongzhou, China.
  • Department of Hepatobiliary and Hernia Surgery, The Third People's Hospital of Yongzhou, Yongzhou, China.
  • Department of Clinical Nutrition, Chongqing Yubei District People's Hospital, Chongqing, China.
  • Department of Clinical Nutrition, The Chenjiaqiao Hospital of Shapingba District of Chongqing, Chongqing, China.
  • Department of Clinical Nutrition, Beibei Affiliated Hospital of Chongqing Medical University, The Ninth People's Hospital of Chongqing, Chongqing, China.

Zitieren

APA

Zhou, J., Zhou, Q., Zhou, L., Wang, J., Gao, J., Qin, L., Tan, B., Fan, P., & Li, C. (2026). Safety, accuracy, empathy, reliability, and readability of large language model chatbot responses to public-facing vegetarian and vegan nutrition advice questions: a cross-sectional comparative study. Frontiers in Public Health, 14. https://doi.org/10.3389/fpubh.2026.1921819

Vancouver

Zhou J, Zhou Q, Zhou L, Wang J, Gao J, Qin L, et al. Safety, accuracy, empathy, reliability, and readability of large language model chatbot responses to public-facing vegetarian and vegan nutrition advice questions: a cross-sectional comparative study. Frontiers in Public Health. 2026;14. doi:10.3389/fpubh.2026.1921819

BibTeX
@article{Zhou2026,
  author = {Zhou, Jiyong and Zhou, Qiongjiao and Zhou, Lijun and Wang, Jiajia and Gao, Jing and Qin, Long and Tan, Bing and Fan, Ping and Li, Changwei},
  title = {Safety, accuracy, empathy, reliability, and readability of large language model chatbot responses to public-facing vegetarian and vegan nutrition advice questions: a cross-sectional comparative study},
  journal = {Frontiers in Public Health},
  year = {2026},
  volume = {14},
  doi = {10.3389/fpubh.2026.1921819}
}
Kostenlos starten →