Young-adult pilot measures GPT-4 counseling acceptability and fidelity
Source check: Source checked. Not independent scientific confirmation.
Finding: In a single-arm pilot,45 young adults interacted with a GPT-4 alcohol-counseling chatbot. Participant-rated relational fidelity was higher in the second phase after prompts were revised (82.6 versus 67.2 on a normalized scale). This is a preliminary delivery/acceptability finding, not evidence of reduced drinking or a causal effect of prompt changes.
Technology: GPT-4 conversational generation in MICA
Family GPT-4: profile · models page
Publication: Peer-reviewed paper · Test context: direct test. Catalog categories, not an evidence grade.
- AI contribution
- GPT-4 generated the counseling conversation delivered to participants, directly enabling the evaluated interaction; model-assisted qualitative coding was an additional role.
- Test and result
- Eight participants in phaseI and37 in phaseII; post-session CEMI and usability ratings, transcript review and descriptive session measures.
- Limitation / still unknown
- Uncontrolled convenience sample and changed prompts; baseline readiness differed. No distal drinking outcomes or causal prompt-effect inference. No inappropriate responses were observed in this small review, which does not establish general safety. Supplementary prompts and exact fine-tuning implementation were not independently inspected.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC12207782/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Actual author-manuscript primary clinical/model Methods, Results and limitations in supported Chrome; exact DOI/front final online date. Author manuscript availability2026 distinguished from journal2025 publication..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 3/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.