Personalizing Structured Self-Help in the Moment: Students’ Experiences With Embedded AI Chatbots in a Guided Digital Mental Health Platform
Abstract
Background: University students can use guided digital mental health interventions without having immediate access to a coach. Fixed programme content does not always answer questions about choosing a programme or applying an exercise to a personal situation. This study examined whether purpose-specific large language model chatbots can address this gap within an existing platform, MoodLift. The study focused on how students judged their usefulness, trustworthiness, and safety. Methods: Two prototypes developed for MoodLift were evaluated: a Program Selection Assistant and an AI Coach. Ten first-year students used both tools. They completed usability and acceptability measures before joining one of two focus groups. Questionnaire results were summarized descriptively. Focus-group transcripts were analyzed using reflexive thematic analysis. Results: Mean Chatbot Usability Questionnaire scores were 70.3 for the Program Selection Assistant and 70.8 for the AI Coach. Acceptability ratings were generally positive. Five themes were developed: (1) from general self-help to personal support; (2) providing direction while preserving autonomy; (3) students trusted the chatbots conditionally, not automatically; (4) embedding made AI more relevant and legitimate; and (5) safety required a bounded, complementary role. Situational fit emerged as a cross-cutting principle: preferences varied between students and depended on the task and context. Conclusions: Purpose-specific chatbots can help students use guided digital self-help at the moment questions arise. Their value depends on reliable responses and an explicit scope. Providers should explain how data are handled and who remains responsible, while ensuring that human support is available
Related
- AI chatbots for health information-seeking and self-diagnosis among Saudi university students: patterns of use, predictors, attitudes, and trust
- Efficacy of artificial intelligence-driven conversational agents (chatbots) for mental health promotion among university students: a systematic review and meta-analysis
- Patients' Experiences in a Dialogue-Based Outpatient Clinic Embedded Within Routine Oncology Care
- Experiences administering the EuroQol five-dimension five-level (EQ-5D-5L) for valuation of health-related quality of life estimates in Kenya: An embedded qualitative study
- Association of Structured Preclinical Phantom Training and OSCE/OSPE-Based Assessment with Endodontic Procedural Complications Among Dental Students: A Retrospective Cohort Study
BibTeX
@misc{struijs2026personalizing,
title = {{Personalizing Structured Self-Help in the Moment: Students’ Experiences With Embedded AI Chatbots in a Guided Digital Mental Health Platform}},
author = {Struijs, S.Y. (Sascha)},
howpublished = {PsyArXiv},
year = {2026},
month = oct,
doi = {10.31234/osf.io/scdb5_v1},
url = {https://osf.io/preprints/psyarxiv/scdb5_v1/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Personalizing Structured Self-Help in the Moment: Students’ Experiences With Embedded AI Chatbots in a Guided Digital Mental Health Platform
AU - Struijs, S.Y. (Sascha)
PY - 2026
DA - 2026/10/30
PB - PsyArXiv
DO - 10.31234/osf.io/scdb5_v1
UR - https://osf.io/preprints/psyarxiv/scdb5_v1/
AB - Background: University students can use guided digital mental health interventions without having immediate access to a coach. Fixed programme content does not always answer questions about choosing a programme or applying an exercise to a personal situation. This study examined whether purpose-specific large language model chatbots can address this gap within an existing platform, MoodLift. The study focused on how students judged their usefulness, trustworthiness, and safety. Methods: Two prototypes developed for MoodLift were evaluated: a Program Selection Assistant and an AI Coach. Ten first-year students used both tools. They completed usability and acceptability measures before joining one of two focus groups. Questionnaire results were summarized descriptively. Focus-group transcripts were analyzed using reflexive thematic analysis. Results: Mean Chatbot Usability Questionnaire scores were 70.3 for the Program Selection Assistant and 70.8 for the AI Coach. Acceptability ratings were generally positive. Five themes were developed: (1) from general self-help to personal support; (2) providing direction while preserving autonomy; (3) students trusted the chatbots conditionally, not automatically; (4) embedding made AI more relevant and legitimate; and (5) safety required a bounded, complementary role. Situational fit emerged as a cross-cutting principle: preferences varied between students and depended on the task and context. Conclusions: Purpose-specific chatbots can help students use guided digital self-help at the moment questions arise. Their value depends on reliable responses and an explicit scope. Providers should explain how data are handled and who remains responsible, while ensuring that human support is available
ER -