September 20, 2025PLoS ONEOpen Access

Performance of large language models ChatGPT and Gemini in child and adolescent psychiatry knowledge assessment

View Full Paper

Ask AI

Bookmark

Authors

Discussion

Assessment of accuracy in large language models ChatGPT and Gemini in child psychiatry, highlighting variations across topics.

Accuracy of the models was 68.3% to 78.9%, indicating generally solid performance in child and adolescent psychiatry.
Statistical comparisons revealed significant differences between Gemini 2.0 Flash and earlier models, showing advancements in capabilities.
Certain topics, like psychopharmacology, proved challenging compared to areas with clear diagnostic criteria such as schizophrenia.
The variability in accuracy highlights risks of misinterpretation and potential biases before clinical implementation.

Neubauer et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

September 20, 2025PLoS ONEOpen Access

View Full Paper

Ask AI

Bookmark

Authors

Assessment of accuracy in large language models ChatGPT and Gemini in child psychiatry, highlighting variations across topics.

Accuracy of the models was 68.3% to 78.9%, indicating generally solid performance in child and adolescent psychiatry.
Statistical comparisons revealed significant differences between Gemini 2.0 Flash and earlier models, showing advancements in capabilities.
Certain topics, like psychopharmacology, proved challenging compared to areas with clear diagnostic criteria such as schizophrenia.
The variability in accuracy highlights risks of misinterpretation and potential biases before clinical implementation.

Neubauer et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark