aceModerate evidence
More
In a survey of 1,872 Chinese students, 14.6% showed high-risk AI use with anxiety and poor sleep
Study title: Problematic generative AI use and mental health risks among Chinese university students: latent profiles, network structure, and health literacy as a modifiable resource
- Evidence type
- Survey
- Sample
- n = 1,872
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2559cite or check it
Study title: Problematic generative AI use and mental health risks among Chinese university students: latent profiles, network structure, and health literacy as a modifiable resource
A cross-sectional survey of Chinese university students at six universities in eastern, central and western China sorted 1,872 valid respondents into four groups. Some 39.9% were low-risk adaptive users, 25.7% anxiety-prone dependent users, 19.8% sleep-disrupted overusers and 14.6% high-risk dysregulated users.
Read the study abstract
Objective Generative artificial intelligence (GenAI) has become a routine academic tool for university students, but dysregulated use may co-occur with anxiety, sleep problems, psychological distress, and impaired academic functioning. This study examined GenAI-related digital behavior as a public health issue and assessed whether health literacy and self-regulated learning were associated with lower-risk profiles. Methods A cross-sectional survey was conducted among Chinese university students from six universi…
The problem
In a survey of 1,872 Chinese university students, 14.6% fell into a high-risk dysregulated group with the worst loss of control over generative AI use, academic anxiety, sleep problems and distress, and 60.1% of them used it daily or almost daily.
In brief
The high-risk group scored highest on loss of control, compulsive use, academic avoidance, academic anxiety, sleep problems and psychological distress, and lowest on health literacy, self-regulated learning and academic engagement. Daily or almost daily use was reported by 60.1% of this group versus 8.7% of low-risk users.
Higher health literacy (OR 2.41) and self-regulated learning (OR 2.86) were linked to greater odds of low-risk rather than high-risk membership. The authors say frequent use alone was not the key signal; the concern was impaired control combined with worry and late-night use. The design cannot show cause and effect.
Main points
- Of 2,140 questionnaires submitted, 1,872 valid responses were analysed after screening removed 268; data were collected April 20 to May 30, 2026.
- Four profiles emerged: low-risk adaptive (39.9%), anxiety-prone dependent (25.7%), sleep-disrupted overusers (19.8%) and high-risk dysregulated (14.6%).
- Problematic AI use correlated with academic anxiety (r = 0.43), sleep problems (r = 0.36) and psychological distress (r = 0.46).
- Health literacy was negatively correlated with problematic AI use (r = −0.31).
- Network analysis identified loss of control, academic worry, delayed bedtime and academic avoidance as central or bridge nodes.
- Credibility evaluation and time management showed negative bridge expected influence, which the authors read as resource-related protective positions.
- Academic engagement was also tied to low-risk membership (OR 1.65), while daily use lowered the odds of low-risk membership (OR 0.38).
The full story
Participants were full-time undergraduates aged 18 or older who had used at least one generative AI tool. The sample had a mean age of 20.31 years and was 59.4% female. Recruitment relied on convenience sampling through course groups, student affairs offices and online class channels. Screening removed 96 incomplete questionnaires, 54 that failed an attention check, 68 completed in under 180 seconds and 50 with patterned responses.
Latent profile analysis used six indicators: loss of control, compulsive use, academic avoidance, academic anxiety, sleep problems and psychological distress. The four-profile solution had entropy of 0.891. A five-profile model created a class of only 4.2% and was judged less stable.
The two middle groups differed in kind. Anxiety-prone dependent users had higher academic anxiety and avoidance, while sleep-disrupted overusers had worse sleep scores. Daily use was concentrated in the high-risk group but also appeared in the two intermediate profiles.
The authors suggest universities fold responsible AI use into campus health promotion rather than rely on blanket prohibition. They frame screening items on loss of control, academic worry, delayed bedtime and avoidance as hypotheses for future longitudinal and intervention studies, not proven interventions.
Sources
1 on file- Frontiers in Public Healthdoi:10.3389/fpubh.2026.1910408
Talk it over
Who carries this cost, and what would have to change to prevent it?
In a survey of 1,872 Chinese students, 14.6% showed high-risk AI use with anxiety and poor sleephttps://truvace.com/item/problematic-generative-ai-use-and-mental-health-risks-among-chinese-university-s
Who carries this cost, and what would have to change to prevent it?
The problem In a survey of 1,872 Chinese university students, 14.6% fell into a high-risk dysregulated group with the worst loss of control over generative AI use, academic anxiety, sleep problems and distress, and 60.1% of them used it daily or almost daily.
Answer in the threadHow do we know? What can this kind of evidence show, and what can’t it?
The evidence This rests on a survey with a sample of 1,872, from Frontiers in Public Health.
Answer in the threadWhat does it leave open? What else would you want to know before acting on it?
The record’s own caveat The cross-sectional design cannot establish causation or temporal order, all measures were self-reported, and the convenience sample of Chinese university students limits generalizability.
Answer in the threadHave you seen AI used this way where you study, work or live? Did it go the same way?
Answer in the thread
The questions are the same on every record. The details under them are taken from this record as written.
What readers say