Study Finds LLM Dependence Patterns Among Nurses

A cross-sectional survey of 459 clinical nurses in six hospitals across three provincial-level administrative regions in China found higher average scores for functional than existential dependence on large language models (LLMs). The study, published in BMC Nursing in 2026, examined factors associated with both types of dependence.

Researchers collected questionnaire data from August to September 2026 using convenience sampling and a structured, validated questionnaire. Mean functional dependence was 35.42 ± 5.81, or 3.94 ± 0.65 per item on a 1–5 scale; mean existential dependence was 27.18 ± 7.06, or 3.02 ± 0.78 per item. The authors said these scores do not establish severity levels or the prevalence of problematic dependence.

In univariate analyses, scores differed significantly by age, educational attainment, frequency of LLM use, and receipt of related training (P < 0.05). Multiple linear regression found that use frequency and related training were associated with both dependence dimensions; educational attainment had a more complex adjusted association. More frequent LLM use was associated with higher scores, while training was associated with lower scores.

The researchers recommended that nursing management address potential dependence risks through standardized training, institutional policies, and ethical oversight to guide cautious and responsible LLM use.

A cross-sectional survey of 459 clinical nurses in six hospitals across three provincial-level administrative regions in China found higher average scores for functional than existential dependence on large language models (LLMs). The study, published in BMC Nursing in 2026, examined factors associated with both types of dependence.

Researchers collected questionnaire data from August to September 2026 using convenience sampling and a structured, validated questionnaire. Mean functional dependence was 35.42 ± 5.81, or 3.94 ± 0.65 per item on a 1–5 scale; mean existential dependence was 27.18 ± 7.06, or 3.02 ± 0.78 per item. The authors said these scores do not establish severity levels or the prevalence of problematic dependence.

In univariate analyses, scores differed significantly by age, educational attainment, frequency of LLM use, and receipt of related training (P < 0.05). Multiple linear regression found that use frequency and related training were associated with both dependence dimensions; educational attainment had a more complex adjusted association. More frequent LLM use was associated with higher scores, while training was associated with lower scores.

The researchers recommended that nursing management address potential dependence risks through standardized training, institutional policies, and ethical oversight to guide cautious and responsible LLM use.