Assessing and alleviating state anxiety in large language models

Journal article

Ziv Ben-Zion, Kristin Witte, A. K. Jagadish, O. Duek, I. Harpaz-Rotem, Marie-Christine Khorsandian, A. Burrer, Erich Seifritz, Philipp Homan, Eric Schulz, Tobias R Spiller
npj Digit. Medicine, 2025

Semantic Scholar DBLP DOI PubMedCentral PubMed

Cite

APA Click to copy
Ben-Zion, Z., Witte, K., Jagadish, A. K., Duek, O., Harpaz-Rotem, I., Khorsandian, M.-C., … Spiller, T. R. (2025). Assessing and alleviating state anxiety in large language models. Npj Digit. Medicine.

Chicago/Turabian Click to copy
Ben-Zion, Ziv, Kristin Witte, A. K. Jagadish, O. Duek, I. Harpaz-Rotem, Marie-Christine Khorsandian, A. Burrer, et al. “Assessing and Alleviating State Anxiety in Large Language Models.” npj Digit. Medicine (2025).

MLA Click to copy
Ben-Zion, Ziv, et al. “Assessing and Alleviating State Anxiety in Large Language Models.” Npj Digit. Medicine, 2025.

BibTeX Click to copy

@article{ziv2025a,
  title = {Assessing and alleviating state anxiety in large language models},
  year = {2025},
  journal = {npj Digit. Medicine},
  author = {Ben-Zion, Ziv and Witte, Kristin and Jagadish, A. K. and Duek, O. and Harpaz-Rotem, I. and Khorsandian, Marie-Christine and Burrer, A. and Seifritz, Erich and Homan, Philipp and Schulz, Eric and Spiller, Tobias R}
}

Abstract

The use of Large Language Models (LLMs) in mental health highlights the need to understand their responses to emotional content. Previous research shows that emotion-inducing prompts can elevate “anxiety” in LLMs, affecting behavior and amplifying biases. Here, we found that traumatic narratives increased Chat-GPT-4’s reported anxiety while mindfulness-based exercises reduced it, though not to baseline. These findings suggest managing LLMs’ “emotional states” can foster safer and more ethical human-AI interactions.