JMIR Publications

ABSTRACT

Background:

Young people increasingly seek sexual health information through digital technologies, particularly when sensitive questions are difficult to discuss with parents, educators, or health care professionals. Conversational artificial intelligence may offer a private, immediate, and accessible way to obtain information through natural-language interactions. However, generative artificial intelligence systems can produce inaccurate, unsupported, biased, or developmentally inappropriate responses, creating important safety concerns for adolescents and young adults. Although conversational agents have been investigated for sexual and reproductive health, qualitative and participatory studies remain among the least represented in this literature and less is known about how diverse youth can be meaningfully engaged throughout the design of generative artificial intelligence systems intended to provide sensitive health information.

Objective:

This study aims to engage youth in the co-design and evaluation of an artificial intelligence–powered chatbot intended to provide safe, reliable, inclusive, and evidence-based comprehensive sexuality education to youth aged 15-24 years in Canada. Specifically, the study will identify youths’ sexual health information and support needs, engage youth in chatbot development, and evaluate the prototype’s usability, acceptability, perceived safety, and utility.

Methods:

This multiphase study is guided by community-based participatory research and informed by intersectionality and user-centered design. In phase 1, a National Youth Advisory Group (N-YAG) of approximately 10-12 youth aged 15-24 years will be established and will remain engaged across all subsequent phases, contributing to study implementation, co-development of the focus group guide, interpretation of findings, chatbot development, usability assessment, and knowledge mobilization. Phase 2 will involve 8-10 virtual focus groups with approximately 80-100 youth aged 15-24 years residing in Canada to explore sexual health information-seeking experiences, unmet needs, barriers to accessing information and support, and expectations of conversational artificial intelligence. In phase 3, qualitative findings and vetted sexual health resources will inform development of a curated knowledge base, which will be validated by a panel of 3-5 content experts using content validity indices and reviewed by the N-YAG, and then integrated into a chatbot prototype using retrieval-augmented generation. In phase 4, approximately 30-60 youth will use the prototype for approximately 2 days and participate in usability testing combining virtual focus groups with standardized rating scales, including the System Usability Scale and the user version of the Mobile Application Rating Scale, alongside study-developed items addressing inclusivity, perceived safety, and trust. Qualitative data will be analyzed using the DEPICT participatory analysis model and managed using NVivo. Rating-scale data will be summarized descriptively, and open-ended responses will be analyzed thematically. The study received ethics approval from the University of Alberta Research Ethics Board (Pro00164574).

Results:

Recruitment is expected to begin in November 2026. Data collection across all 4 phases is anticipated to conclude in March 2027. Initial study findings are expected to be disseminated and submitted for publication in fall 2027.

Conclusions:

This study places young people within the design and interpretation processes of an artificial intelligence–supported sexual health intervention. Integrating participatory research with a validated sexual health knowledge base and retrieval-augmented generation may provide practical insights into the development of safer and more inclusive conversational artificial intelligence for developmentally sensitive health information. Clinical Trial: NA