This is an ongoing study and overall findings are subject to changes.
When I conducted the overarching chatbot study with the general young adult population, I realized that the health needs of LGBTQ+ individuals were unique. Immediately, I started questioning how this would make the user experience for LGBTQ+ individuals seeking sexual health information on chatbots different from other users.
I was curious about the sexual health questions they had and if the chatbot could provide suitable answers for their specific questions. Hence, to take a deep dive into these questions, I conducted a semi-structured interview followed by an unmoderated usability test using the Lookback software.
Before showcasing some of the unique findings from this subgroup analysis, I wanted to ascertain that through the use of Lookback, I was able to find specific examples where users were facing challenges when using the chatbot. Video recordings allowed me to create highlight reels showcasing these challenges clearly. I think this visual aid would be very useful when communicating feedback to designers on the team.
Findings
While participating in the usability test, participants were encouraged to use the "Think Aloud" method, voicing their thoughts while they concurrently accessed the chatbot. This allowed me to further capture their candid thoughts at the initial point of interaction (which are often authentic feedback). These perspectives could be very useful for future improvements.
For privacy and confidentiality purposes, I have removed all identifying characteristics of participants and will only use screenshots to visually support some findings.
The prompts were useful! In the early stages of interacting with the chatbot, some LGBTQ+ participants mentioned that they were not sure of the language to use or the exact type of questions they wanted to ask. Participants who were in the initial stages of coming out felt that they "did not know how to use some terms correctly."
As such, the prompts provided by the chatbot were very helpful in accessing the questions and being signposted to the correct information. More than half of the participants clicked on "Sure" when asked if they wanted more information for their question.
2. Frustration- Chatbot does not get me!
Multiple participants had trouble seeking accurate answers for their concerns as LGBTQ+ individuals. Below are some authentic questions that could not be answered by the chatbot. By using the "Think Aloud" method, I was able to analyze the participant's frustration in their tone too.
In the image below, a participant had their cursor on the "Ask the Experts" option.
When they clicked it and typed in "Gay" as a keyword, the outcome did not provide relevant or suitable information for the information they sought. Instead, they were provided with a variety of information that could be confusing for some. Below is a screenshot of that example.
This participant decided to type a more specific key word ("Gay Sex"), to which inaccurate information was shown again. They mentioned verbally "Wait... this is definitely not what I was looking for."
The participant keyed in "Men Gay Sex" on the third try and unfortunately, did not receive accurate information. Instead, similar "exploratory/general" information was repeated. They mentioned that speaking to a virtual agent could have been less frustrating.
3. Can Roo be more personable?
The last finding from the initial round of testing was about the design of the chatbot. Participants felt that if the chatbot looked more like them, it would be more relatable. This was a common perspective shared across multiple participants in the semi-structured interview component. Using avatars or even AI-powered virtual human beings similar to users could increase the adoption and use of the product. Below is an example of AI-powered human avatars ("Artificial Humans" ) designed by Samsung.
As the study continues, I hope to be able to provide more useful strategies to improve the user experience for LGBTQ+ individuals on this platform. With the necessary improvements, alongside more accurate and relevant health information suitable for LGBTQ+ individuals, I believe this could improve the overall health information-seeking experience for the LGBTQ+ community.
Commenti