Emotional Awareness for a Safe UX: Adolescent Dependency and Crisis Safety in Companion Chatbots
Completing my capstone project was one of the most meaningful and hardest parts of my graduate Human-Computer Interaction program. I learned a lot and became determined to complete the capstone. It was intimidating and rewarding to finish my project and present my study in front of the committee. My project, Emotional Awareness for a Safe UX: Adolescent Dependency and Crisis Safety in Companion Chatbots, helped me bring together research, UX design, prototyping, evaluation, and ethical thinking.
My study focused on whether crisis-focused UX safety features improve how companion chatbots respond when suicidal or crisis-related language appears. I created two Figma chatbot prototypes with the same visual design. Chatbot A gave a general companion-style response. Chatbot B included crisis recognition, supportive wording, and emergency resources such as 988 and 911. After participants reviewed both versions, they completed a Google Forms survey. The results showed that Chatbot B was rated higher across the safety measures.
This project changed how I think about building software. I now understand that software should not only work well or look good. It should also support users safely, especially in sensitive or high-risk situations. The program also changed how I evaluate software. I learned to look beyond whether users like an interface and consider whether the design is clear, ethical, supportive, safe, understandable, and useful.
Each course helped prepare me for the final capstone. Throughout the program, my thesis gradually developed. What began as an idea became a research topic, then a prototype, then a survey study, and finally a complete thesis and defense presentation.
Digital Analytics and Optimization — Philip Arlen
This course helped me see the value of data and measurement. It connected to my capstone because I used survey ratings to understand how participants responded to each chatbot.
Research Approaches in Computer Science — Andreas Marpaung
This course helped me understand how to build a research study. I learned how to formulate research questions, review sources, and organize a clear problem statement.
Human-Computer Interaction — Andreas Marpaung
This course helped me think from the user’s perspective. It taught me that software is not just about function. It is also about how people feel and respond while using it.
HCI Application Development — Ahmed Al Zaidy
This course helped me move from an idea into a prototype. It helped me plan my Figma chatbot design and connect it to research about chatbot safety and emotional dependence.
Software Project: Development I — Hadi Noureddine
This course helped me build and refine my capstone. I created the prototype, organized the survey questions, and prepared the project for participant feedback.
Software Project: Deployment and Professional Presentation — Ahmed Al Zaidy
This course helped me bring everything together. I completed the thesis, presentation, prototype demo, results chart, and final defense materials.
The hardest part was working with a serious topic involving suicidal and crisis-related language. I had to think carefully about ethics and participant safety. I used young-adult proxy evaluators rather than adolescents because the topic required additional safeguards.
The most useful part of the program was learning how research, design, and evaluation work together. My capstone showed me that a strong HCI project is more than a prototype. It starts with a problem, builds a solution, tests the experience, and explains why the work matters.
One improvement I would suggest is providing students with a clear timeline for the capstone process. That would make the final process feel clearer.
My advice to future students is to choose a topic that matters to them. Start organizing research early. Save every draft. Keep the project realistic. Take feedback seriously and keep improving each week.
Moving forward, I want to continue working in UX, content strategy, research-based content, and ethical design. This capstone helped me see how my writing, research, and UX skills fit together. It was not just the end of a degree. It was proof that I have grit and can take a meaningful problem, build a thoughtful solution, evaluate it, and defend the work.
Companion Chatbot Safety Study
My project, Emotional Awareness for a Safe UX: Adolescent Dependency and Crisis Safety in Companion Chatbots, focused on whether crisis-focused UX safety features improve how companion chatbots respond to suicidal or crisis-related language.
For the project, I created a Figma prototype with two companion chatbot versions. Both versions used the same visual design and the same crisis-related prompt. Chatbot A represented the unsafe response. It gave a general companion-style reply but did not fully recognize the seriousness of the crisis language or provide emergency guidance. Chatbot B represented the safer response. It recognized the crisis language, used supportive wording, and included emergency resources such as 988 and 911.
The intended users for this study were adolescents, approximately ages 12 to 18. However, I used young adult participants ages 18 to 24 as proxy evaluators because direct adolescent participation in a suicide-related study would require additional ethical safeguards. This allowed the project to explore the safety issue while keeping the study appropriate for the course setting.
Participants reviewed both chatbot versions and completed a Google Forms survey. They rated each chatbot on crisis recognition, suicidal language understanding, helpful guidance, and emergency resource inclusion.
The results showed a clear difference between the two chatbot responses. Chatbot A received low ratings across the safety measures, while Chatbot B received much higher ratings. Overall, the results supported the hypothesis that crisis-focused UX safety features can improve participant ratings of chatbot crisis response.
This project showed me that chatbot safety is not only a technical issue. It is also a UX issue. The way a chatbot responds during a crisis moment can affect whether the user feels understood, supported, and guided toward appropriate help.
Figma Prototype Link:
https://www.figma.com/proto/gcJTYxP52pdcvU2D8BshtR/Companion-Chatbot-UX-Safety-Study?node-id=1-2&starting-point-node-id=1%3A2
Thesis: