Design For Tension

In this project, we wrote the script for a chatbot in Flow XO that was designed to educate the user on a controversial matter in a human fashion. The topic we chose was gun violence and laws.

We decided our chatbot should be fair and try to convince the user of the opposite of their standing viewpoint, or at least expose them to information from the other side. We researched facts for both sides. Two of us researched facts about gun violence in terms of statistics, while the other two researched in-place laws on gun control. I researched already existing laws on gun control in opposed to gun control, meaning the narrative I built was that sufficient gun regulation was already in place. I was additionally tasked with writing those facts into the Flow and combining all the Flows into one.

Our main design question was how to allow the user to interact with the bot, as we wanted to avoid a bot that spit out all the facts it had. Our main question was how the user can respond to the facts presented by the bot. Our initial idea was to allow the user to respond to each fact with “Agree”, “Disagree”, “Tell me another fact”. Another more sophisticated idea was to allow the user to rebut the bot’s facts with facts from the taken from the other side of the debate. For instance, if the bot gave a fact such as “x% of youth fatalities are from gun violence”, the user would have the optional to respond with “many gun violence statistics are skewed by counting suicides”, “those guns are obtained illegally as minors cannot purchase a firearm”, etc. This would have made for a more natural conversation but would have made the process an order of magnitude more difficult. It additionally leads to the possibility of dead-end or cyclical conversations.

In the end, each flow had a slightly different way of handling responses. One stuck with the “Agree”, “Disagree”, “Tell me another fact” format. Another used a similar format but tailored each response to be contextual with the fact given. I eliminated the “Agree” and “Disagree” responses, and instead the user could ask for another fact based on subject (in this case, state or federal law).

A Demonstration of all possibilities: