Design for Tension: Medicare for All
My group has 4 members: Grant, Ben, Matthew, and me and our chat bot topic is “Medicare for All”. We created our chat bot using flowXO. I was in charge of creating the flow and working with flowXO. My role in the project was creating the flow and making sure the chat bot is functional. Matthew tested the paper prototype with his friends and recorded the feedback. Ben made the demo video. Grant did research on the topic. We tried to create a chat bot that helps people better understand “Medicare for All” and point people to new sources they might be unlikely to see.
Brainstorming & Sketching
Before the break, we brainstormed and created the paper prototype.
This is the paper flow we created before the break. The beginning part of the flow asks the user if he/she knows what “Medicare for All” is. It provides a decent amount of introduction if the user is not familiar with our topic. It will lead the user to the next part of the flow otherwise. The middle part of the flow mainly discuss “Medicare for All” with user. It first asks user if he/she supports the plan and the reason behind. Then it provides some opposing thoughts to prompt the user to think differently. In the end, it asks if user changed his/her mind and leads the user to join Bernie, get in touch with us, or just leave the chat bot.
The beginning part of this flow is designed for people who are not familiar with current issues and American politics. I thought it would be good to provide user with some info before talking about it because I personally have never heard about it before. Since we wanted to point user toward ideas they might be unlikely to see, we have to first ask the user’s opinion on “Medicare for All”. Then we can try to inspire the user with some contrasting thoughts and see if we can slightly alter what the user thinks. Whether if we can persuade the user or not, we will prompt the user to either get more involved with “Medicare for All” or talk more with us. But of course, we can let the user leave the chat bot if he/she is not interested.
Basically, the beginning part of the flow needs to be optimized, the discussion needs more loops to feel like a discussion than a survey, and the ending needs to be more smooth.
We met together after break and made a final sketch of the flow.
We changed the wordings to be more engaging and tried to be less annoying. For example, we asked “What do you think?” instead of “Do you support it?”. We changed the ending to a “thank you” message instead of prompting the user to join Bernie because that was abrupt. Most importantly, we modified the discussion part to have different aspects instead of just good or bad. We also sketched a loop in the end to get the user back to discussion if he/she wants to keep talking.
Then I implemented our whiteboard flow into flowXO.
As you can see, our chat bot has 3 phases and the user can choose what to ask about “Medicare for All” if he/she is not familiar with it.
After I created the chat bot, we brought it to class on demo day.
We let the users play around with it first and asked if they have any questions or concerns. Then we showed them the flowXO and talked to them about our train of thought. In the end, we asked them to give feedback and recorded their words.
We got a lot of useful feedback from demo day. For example, one of the user said he has heard about “Medicare for All” but regrets for not asking for introduction when the bot asked his opinion about “Medicare for All”.
So I improved the flow by adding an extra question. The user can only skip the beginning phase and go straight to the discussion phase if he/she chooses “I’m very familiar”. Otherwise, the user will be asked if he/she would like an introduction.
One of the user said “What if I just don’t care?” when the chat bot asked for her opinion on “Medicare for All”.
So I added a feature to stop the rest of the flow when the user is simply not interested. I think it is fairly reasonable to give user the freedom to not enter the discussion.
The work that I completed comparing to the rest of the group is more about functionality of our chat bot. I was the only person managing flowXO since it doesn’t allow users to share a project. For example, our discussion phase requires a lot of content since we wanted to provide the user with opposing ideas. Unfortunately, I grew up in China so I’m not familiar with any American politics at all. So my teammates wrote the content of the discussion phase.
This is the part that I’ve done before meeting with the team. Since I was in charge of flowXO, I created the flow first and made sure it works the way we wanted. Then we met to complete it as a group.
This is what the conversation looks like after filling in the content of discussion.
I spent a long time to word the ending phase carefully and implement the loop back to discussion.
If the user would like to explore other aspects of “Medicare for All”, it will take the user back to the discussion phase so the user can play around with our bot without having to restart the conversation.
From my point of view, it wasn’t easy to word and structure the ending phase. I didn’t want the user to feel offended or annoyed about whether he/she changed mind and the bot prompting he/she to talk more. Also, I thought we needed a clear message to let the user know that this is the end of the flow.
Our chat bot is designed for people who care about current issues in America, especially “Medicare for All”. The introduction phase of our chat bot also serves to include people who don’t know about “Medicare for all” in our user group. The overall goal of our application is to prompt user to explore “Medicare for All” from various perspectives and hopefully challenge their original belief.
The user feedback was extremely important for this project. The ultimate goal for a chat bot is to be very conversational like a real person but make sure the user comfortable at the same time. It was impossible for us to just design a chat bot by imagining the possible user responses. As mentioned above, our interaction with users has caused numerous improvements. So I guess the more user testing we do the comprehensive our chat bot can be.