Design For Tension
The goal of this project was to create a conversational chatbot with the purpose of convincing the user of a new idea they may have not encountered before. We picked alternative energies, a consistent topic of contention today. Our goal was to gage the users level of knowledge about alternative energy and determine their current stance on the topic. The bot then uses this information to choose the best information to present to the user.
We started by researching the topic and gathering what information we wanted to present to users. We choose our own stance and made plans to make the bot conversational and friendly to users. We prototyped some small conversation and then moved into our full prototype for demo day.
We discussed several options for what we wanted our chatbot to convey. In the end we settled on alternative energies. We quickly moved on to brainstorming what kinds of energies we wanted to discuss with our bot. We initially researched, natural gas, geothermal, oil, wind, solar, tidal, nuclear, and biofuel. Although we initially wanted to discuss all of these forms of energy we were later limited by the scope of the prototyping tool.
We spent some time discussing how the conversation should flow. We felt that the bot should gage the users interest and stance on the topic and then move forward from there. The goal, if the user was not interested in the topic, was to give them more information and then ask again if they wanted to know more. In this way we wanted to attempt to convince users to learn more and change their mind about the subject.
We used FlowXO to create our bot. This tool ended up forcing us to limit our bot and focus down on the specific message we wanted to convey. We wanted to make the users interaction with the bot short and friendly. We felt the bot would be most useful if the users felt it was non-threatening and wanted to help them. In my group everyone participated in making the bot. I was responsible for much of the initial programming of the bot. I learned a lot about how to use FlowXO and conveyed that knowledge to my team so that we could all make small parts of the bot.
Demo day went well for our bot. It was tested by several of our peers and we received good feedback on our prototype. Testers pointed out some looping issues with our bot and found some glitches. They commented on how natural the length of each message from the bot felt overall, although in a few areas they wanted messages to be broken up. Sometimes the bot felt conversational but sometimes the user felt like they were just browsing through information.
Other testers disagreed though and felt that their own responses had personality and that this added to the conversational nature. In general it was not a very persuasive bot, but testers liked the way information was presented and felt that the information they got from the bot was accurate and interesting.
Our bot does a good job conveying relevant information in a bite sized format. It does not flood the user with information and keeps each message fairly short. If I were to do this project again I would limit the scope of the topic more from the beginning and focus more on the conversational aspects of the bot. I would also spend more time on how we wanted to present the information. I think more time