JMIR Publications

ABSTRACT

Background:

In a society where wellbeing issues are increasingly common, how can we sustainably enhance wellbeing? While various wellbeing interventions (WBIs) are available across different contexts (eg, workplaces, schools, communities), their effects tend to be modest and short-lived. A recently developed theory suggests that the limited impact of these interventions may stem from insufficient focus on fostering wellbeing literacy (WL)—the ability to understand and use wellbeing language effectively across contexts to improve one's or others' WB. WL may enhance the effectiveness of WBIs. However, research on interventions and tools to increase WL remains limited. Developing WL is an interdisciplinary challenge that requires insights from diverse fields (eg, education, psychology, linguistics, sociology, computational science). Given WL's language focus, recent advancements in computational science, such as natural language processing, could be groundbreaking for promoting WL and thus potentially improving established WBIs’ efficacy.

Objective:

This project aims to enhance the effectiveness of WBIs by developing an AI-based intervention tool that fosters WL interactively. To be effective, the tool must adapt to diverse users' contexts and needs. The AI would provide WL-related content (eg, interactive text) tailored to users' WL levels. The objectives are to: 1) Co-design an AI chatbot guide to improve WL in diverse adult users; 2) Evaluate the tool’s acceptance, facilitating conditions and obstacles; 3) Test whether using this tool before a WBI (ie, the “Three Good Things” exercise) enhances the intervention's outcomes.

Methods:

1a) A co-design process is followed in which potential users of diverse social backgrounds and stakeholders with relevant expertise meet multiple times to share their experiential, professional and academic knowledge relative to WL and WBIs, as well as coaching and AI technology, with the goal of informing the design of the chatbot. Building on their insights and a review of the WL literature, an interdisciplinary research team from mathematical and computational sciences, health, psychological and social sciences is developing the prompts that will guide the chatbot. 1b) A mixed-methods approach will be used to evaluate the tool's acceptance and feasibility, including quantitative assessments and qualitative user feedback. 2) A trial will test whether the AI tool improves a WBI’s efficacy.

Results:

Funding has been acquired as of April 2025. Phases 1a and 1b are expected to conclude by November 2026 and September 2027, respectively. Phase 2 is expected to conclude by November 2027. Outputs are expected to be disseminated by February 2028.

Conclusions:

The project aims to provide a proof of concept for the first-ever AI-based tool intended to improve WL. Integrating WL, AI, and state-of-the-art human–computer interaction features (ie, engagement, personalization) could have a game-changing impact on wellbeing promotion, helping people from diverse backgrounds be better prepared to undertake WBIs and optimize benefits.