AI Caramba: The Experience of Hispanic Mothers with Brat, a Parenting Intervention-Enhancing AI-Chatbot.

Background and Purpose: Parent training interventions are recognized as the golden standard for preventing and addressing youth mental health, particularly among underserved families. Regardless, participation is often hindered by time, transportation and logistic constraints making virtual implementations more accessible and popular. Virtual Parent Training Interventions (VPTIs), however, usually replicate in-person manuals and fail to leverage technological capital for personalization, cultural tailoring, practice and/or feedback. The inclusion of chatbots in medical and mental health fields has been positive, which represents new opportunities to enhance VPTIs by offering a highly customizable and easily scalable tool for rehearsal and guidance, while increasing participation and accessibility for underserved populations. Guided by implementation science and a commitment to social justice, this exploratory study examined the acceptability and feasibility of BRAT (Boundaries, Resilience, and Affection Training), a text-based, chatbot designed to simulate adolescent responses and provide PLH component-based guidance (i.e., Positive parenting, nonviolent discipline, collaborative problem solving and engagement). To inform its addition into future intervention trials, this study aimed to learn from the experience of a group of Hispanic mothers that interacted with BRAT.

Methods: Fifteen Hispanic mothers of adolescents (x̄ = 44.53, s = 9.45) from Central Texas participated in a 60-minute virtual session where they customized BRAT using focal adolescent’s sex and age, then engaged in three scenarios: (1) Homework resistance, (2) Excessive screen time, and (3) Disrespectful behavior. Scenarios lasted approximately 10 minutes and were facilitated by a team member who read aloud BRAT’s responses. BRAT generated adolescent responses varying in tone and difficulty (i.e., Negotiation, procrastination, sarcasm, and escalation), while incorporating idioms, ethnicity-specific expressions and behaviors. Following the scenarios, participants completed semi-structured interviews exploring usability, realism, feedback, utility, acceptability, and perceived value. Transcripts were analyzed using reflexive thematic analysis, with iterative coding and team-based consensus to enhance analytic rigor.

Results: Three themes emerged: (1) Realism and accuracy, (2) Feedback and self-reflection, as well as (3) Potential uses and promising avenues. Mothers reported having an overall positive experience interacting with BRAT as it closely mimicked the type and quality of responses from their adolescents. BRAT was deemed culturally accurate due to mixing languages and using idioms, while feedback using PLH components was perceived as useful, eye-opening, and delivered in a friendly way. Participants saw BRAT’s potential for scenarios regarding substance use, pubertal development and sexuality.

Conclusions and Implications: Participating mothers enjoyed interacting with BRAT, learned from the experience and receiving feedback, which signals BRAT’s acceptability while underscoring its potential as a successful treatment enhancer within VPTI. By reducing participation barriers and offering opportunities to explore interactions in a tailored platform with evidence-based guidance, BRAT illustrates how chatbots can reinforce access to parent training interventions. As a scalable and adaptable tool, BRAT and other chatbots may also help strengthen engagement, retention and even clinical efficacy of VPTIs. Future research should examine BRAT’s integration into synchronous and asynchronous VPTIs and evaluate its impact on engagement as well as parenting outcomes in trials.