Connecting people through chatbot filtering

The lockdown has continued to isolate students from their old social life. The mental health state of many is dwindling significantly compared to pre-corona times. When we are unable to meet interesting people, what do we do?
With that as a leading question, a chatbot was created to filter individuals on their interests. When we cannot meet offline, we will have to meet online. The prototype saw a good fit with the users, albeit some necessary changes to better fit the narrative of a social hub.
With the limitations of basic chatbots, there was a need for a creative new solution. Compared to the initial idea of having a chatbot as a conversational partner, the chatbot was better fit to connect people in the form of a filter.

Premise

Creating a socialising hub for students during the pandemic through chatbot filtering.

The Problem

In the current pandemic the worries about the mental wellbeing of young adults is an almost daily topic. However, nothing seems to be done about their isolated state. Social anxiety is part of the problem for people during the pandemic (Panchal et al, 2021). Many people have limited contact with the outside world with ongoing lockdowns.
Especially among the student demographic, the consequences of a lack of social life is felt. As a student myself I am feeling the consequences as well. With no social outlet other than my housemates, my social life has come to a stop.

The Beginning

The initial idea for the prototype was a chatbot created with IBM Watson. The user would be able to have natural conversations with the chatbot in order to feel less lonely. The chatbot would be less chatbot and more of a listening ear for socially isolated individuals.
The first step before the design process was getting familiar with the IBM Watson software and its workflow. The second step is drawing out a conversation flow. A conversation flow is similar to a wireframe, it visually shows what the design of the conversation will look like. The first conversation flow was meant to divide the conversation in three distinct emotional states, bad, neutral and positive. The chatbot would then be able to discuss the reason for the emotional state with the individual. However, the variation of possible causes was astronomic. The conversation would be impossible to design without any machine learning and large datasets.

Early on it also became clear that the options with IBM Watson were limited. IBM Watson would not be able to have an entire, natural conversation with a human being. Perhaps a different software was needed together with a different approach to solving the issue of social isolation.

The Process

After IBM Watson proved to be insufficient for the desired prototype, I looked into Power Virtual Assistant. Power VA is a software/web application created by Microsoft to create chatbots as intricate as you would like. The interface and creation of a chatbot is intuitive. I created the second iteration of the conversation flow, this time with a different goal in mind.

Since the goal is still to allow people to socialise, the chatbot will filter individuals based o

n a couple of questions. The chatbot will provide a link to a private chatroom where individuals can gather to chat about the topic of their choice. All chat rooms are connected to broad topics, for example different hobbies. The filtering of the chatbot ensures a degree of guarantee that other people in the chatroom are there for the same reason and have an interest in the topic. The chatbot is now less of a conversationalist and more of a hub to connect like minded individuals.

The chatbot in its form during the new iteration was still aimed at battling loneliness. The tone of voice of the text was largely aimed at providing support and acknowledging the situation the individuals were in.

The Testing

Two students were the subjects for the testing of the prototype. They were given the goal to get to the chatroom of their respective interest. The feedback was mostly regarding small adjustments and the use of a different tone of voice. As the current tone of voice was regarded as condescending to people in a difficult situation. Below we find the notes on the user testing. As all notes are in Dutch, a short summary is given in English below.

General feedback was that the chatbot did not feel like a “chatbot against loneliness”, but rather as a social hub. One compared the chatbot to Tinder. The social hub idea was leading for the last iteration, since it resonated with the solution and problem. As mentioned before, the text has also been rewritten for a tone of voice better fitting the idea of a social hub. Additionally, the questions about loneliness were disregarded. The reason being that a social hub does not need those questions.

Another piece of feedback was the static feeling of the conversation, since the options are clickable and not typable. The feedback could be used in a future adaptation of the prototype. The reason for the clickable options being the difficulty in providing all the different variations of certain responses.
Lastly, the text was too long. In some instances it would show in a messy way, the text was made shorter and more to the point. Additionally, users would prefer to have the options vertically stacked instead of horizontally shown. Sadly, the stacking of options could not be altered with the basic version of Power VA.

Both participants noted their preference of being able to socialise with other people instead of solely talking to a bot. As they said, “a bot acting as a human is off putting.”

Final Iteration

The final iteration of the solution is a more focused and narrowed down version. With a better fitting tone of voice and to the point text. It is now a small filtering hub for students looking to hang out with like minded people.

Link to the prototype: SocialBotHub

An important addition is the Discord channel. Private channels were created for each topic of conversation, making sure that people have solid experience while chatting with each other. Discord was the best option for the chatrooms, as it allows the free creation and monitoring of the chatroom. Additionally, the rooms can be made private, only individuals with a link can join the rooms.

As can be seen from the conversation flow, the chatbot functions as a funnel or decision tree.

 

Reflection

The project was fun to create. The users saw real potential in a platform that connects people through a chatbot. The platform is intuitive in its interaction since all interaction is based on language. One participant of the user testing even brainstormed with me to see what a fully realized version would look like. Stating that the possibility of going to organized events with the people from the chatroom would be amazing.

The chatbot for now is limited to clicking, which was not the initial idea. However, one major lesson that I have learned is that most common applications of chatbots or conversation agents are still narrow. The pre-made options were the only realistic outcome. When you take the limitations of chatbots in consideration and apply them creatively, lots of solutions are still possible to make. What started as an independent conversation agent for lonely people, turned out as a social hub. Which does make sense when I reflect on the project, nothing beats real human social interaction.

 

Recommendation

For future references, the prototype has the potential to grow to a large scale. The social hub can encompass plenty more hobbies and other interests. For now the chatbot is only a filter, but future prototypes could entertain the idea of the chatbot recommending certain chat rooms based on user preferences.

 

Conclusion

In conclusion, the project was a great lesson on working around the limitations of a specific technology. Power VA is a solid platform to create simple chatbots and if the limitations are clear, there are a lot of possible solutions that can be created.
The fight against loneliness is still ongoing and will not be won through this prototype alone. The chatbot is not able to have human-like conversation with a human, but it is great at connecting people with similar interests. The end prototype better fitted the users than the initial idea of a true chatting bot.

 

References

Hudson, J., Ungar, R., Albright, L., Tkatch, R., Schaeffer, J., & Wicker, E. R. (2020). Robotic Pet Use Among Community-Dwelling Older Adults. The Journals of Gerontology: Series B, 75(9), 2018-2028.

Panchal, N., Kamal, R., Orgera, K., Cox, C., Garfield, R., Hamel, L., & Chidambaram, P. (2020). The implications of COVID-19 for mental health and substance use. Kaiser family foundation.

Leave a Reply

Your email address will not be published. Required fields are marked *