Beyond English: ChatGPT's instructions across EU languages
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60460709%3A41110%2F24%3A100014" target="_blank" >RIV/60460709:41110/24:100014 - isvavai.cz</a>
Result on the web
<a href="https://doi.org/10.1016/j.resuscitation.2024.110367" target="_blank" >https://doi.org/10.1016/j.resuscitation.2024.110367</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1016/j.resuscitation.2024.110367" target="_blank" >10.1016/j.resuscitation.2024.110367</a>
Alternative languages
Result language
angličtina
Original language name
Beyond English: ChatGPT's instructions across EU languages
Original language description
This study evaluates ChatGPT's ability to provide cardiopulmonary resuscitation (CPR) instructions across 24 official European Union languages. Using GPT-4o, three responses were generated per language and evaluated by three large language models on medical accuracy, comprehensiveness, clarity, safety, and emergency suitability. While English, Dutch, and Polish demonstrated optimal performance, and some languages like Greek showed more suboptimal ratings, no instructions were rated as critically unsafe. The findings suggest consistent baseline reliability across languages while highlighting the need for comprehensive multilingual benchmarks in healthcare AI systems.
Czech name
—
Czech description
—
Classification
Type
J<sub>imp</sub> - Article in a specialist periodical, which is included in the Web of Science database
CEP classification
—
OECD FORD branch
30221 - Critical care medicine and Emergency medicine
Result continuities
Project
—
Continuities
S - Specificky vyzkum na vysokych skolach
Others
Publication year
2024
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Name of the periodical
RESUSCITATION
ISSN
0300-9572
e-ISSN
0300-9572
Volume of the periodical
202
Issue of the periodical within the volume
110367
Country of publishing house
CZ - CZECH REPUBLIC
Number of pages
2
Pages from-to
—
UT code for WoS article
001310334700001
EID of the result in the Scopus database
2-s2.0-85202652226