Premise
A chatbot that can make sharing feelings of loneliness easier for young adults by making it more fun.
[aesop_parallax height=”50px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Synopsis
With the rise of COVID-19 and it’s restrictions, loneliness among young adults caught my attention. The reason this topic caught my attention is because I am one of those young adults that is affected by the COVID-19 pandemic and can therefore sometimes experience feelings of loneliness. However, I have a group of other (young) adults were I can share my story with, a lot of young adults have no one to talk to or are afraid to talk about their feelings, due to a stigma on loneliness. A chatbot can therefore be a good way to help young adults talk about their feelings of loneliness, without being afraid of what others might think. By experimenting and testing with different chatbots, a chatbot has been created to help young adults express their feelings of loneliness.
[aesop_parallax height=”50px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Substantiation
Since the arrival and the continued impact of the COVID-19 pandemic, and it’s restrictions, loneliness has become a more widely recognized critical public health problem (Killgore et al., 2020). Particularly among young adults (18-24). Young adults have a high-risk for developing feelings of loneliness (McQuaid et al., 2021) because of the instability of their social networks, related to changes in school, identity exploration, or physical changes that can make young people vulnerable to exclusion (Qualter et al., 2015). Loneliness among these young adults is often stigmatized because of the social pressures to appear connected (Pitman et al., 2018). Because of this stigma en social pressure, young adults often ignore, hide and/or trivialize their true feelings (Cacioppo & Cacioppo, 2018). Further, young adults also say that they do not always have or trust an adult to go to for advice and support when they experience these feelings. This can result in people avoiding social situations and experience making meaningful connection with others, very hard. But it is precisely making connections that is important when fighting against feelings of loneliness.
[aesop_parallax height=”10px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
That is why I want to investigate how chatbots can make sharing feelings of loneliness easier and more fun among young adults. The reason I want to prototype and experiment with chatbots is because, they have the potential to be useful tools for people with mental disorders, especially those who are afraid to seek mental health advice due to stigmatization (Abd-alrazaq et al., 2019). Since chatbots do not think and cannot form their own judgments, people feel more comfortable confiding in them without fear of being judged (Lucas et al., 2014).
[aesop_parallax height=”50px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Prototypes / experiments
First, I started my research with developing more knowledge about chatbots and how to make one myself. This was my first step, because I never had created or worked with a chatbot before. While doing desk- and literature research into chatbots, I stumbled upon ‘Landbot.io’. This was an easy tool to quickly create knowledge on how to create a rather basic chatbot. With this tool I decided to create a simple chatbot and perform a quick test. I wanted to investigate whether a chatbot actually has a different result compared to for example a survey, when it comes to young adults sharing feelings.
[aesop_parallax height=”30px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Test 1
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
What
For the first test I decided to use the Myers-Briggs Type (MBTI) Indicator test (Myers, 1962), also known as the 16 personalities test. I made 3 young adults (2 who experience feelings of loneliness, and 1 who has experienced feelings of loneliness) fill out the online personality test. Next I made them go through a chatbot I created. This chatbot was based on the same MBTI test, only I made it shorter and ‘simpler’ in questions (figure 1).
[aesop_parallax height=”30px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Figure 1:
[aesop_video src=”youtube” id=”VQiVz0IRTsU” width=”100%” align=”left” disable_for_mobile=”on” loop=”on” controls=”on” mute=”off” autoplay=”off” viewstart=”off” viewend=”off” show_subtitles=”off” revealfx=”off” overlay_revealfx=”off”]
[aesop_parallax height=”30px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Why
The reason I used the MBTI test as a starting point for this test is because this test contains very personal questions, that users might find hard to fill out. Therefore, this test could give me an insight whether a rather basic chatbot would already have a different impact on the respondents. Further, creating this chatbot helped me created more knowledge on this technology.
[aesop_parallax height=”30px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Results
- The online MBTI form was experienced as exhausting and a bit difficult due to the amount of questions the form contained;
- The chatbot was described as ‘short but sweet’ and easy to fill out in between activities;
- The chatbot was also experienced as more fun because of the different way of interacting. One respondent said: “The chatbot, I like better, because is feels like I have a short conversation. The other one is just filling out a lot of questions”.
- Finally, the respondents experienced the tests as confronting (especially the MBTI form one) and fun because of the reward in the end: the outcome that tells them what personality they are.
[aesop_parallax height=”30px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Conclusion and Reflection
Earning a reward plays a big role when it comes to creating a more fun and easy to ‘talk’ to chatbot. Further, the chatbot does have a different effect on young adults interaction compared to the form. However, it is likely that the fact that the number of questions in the chatbot was a lot less than in the form, influenced the outcome that the chatbot was ‘more fun’.
[aesop_parallax height=”50px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Test 2 : An A/B uhm and C? test
[aesop_parallax height=”10px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Why
Because the BMTI questions are not directly related to talking about feelings of loneliness, they may have been perceived by the respondents as less loaded and therefore more ‘fun’. In my next test I wanted to test this assumption by using questions from the UCLA loneliness scale (Russell et al., 1980). Based on the answer to these questions, loneliness can be detected. Further, I wanted to test what elements of a chatbot could have an influence on the user experience of the young adults, while interacting with it. So in other words, what elements can be used to make a chatbot more fun and accessible.
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
What
I created three chatbots with Landbot.io based on the UCLA loneliness scale questions. The first chatbot consisted out of a more playful colorscheme (figure 2). The second chatbot consisted out of a colorscheme that was more responsive to users trust (figure 3). I wanted to test what the effect of a different visual appearance had on the respondents. The third chatbot consisted of added GIF’s (figure 4). This was based on the theory that images-based interactions can confer to their users to a decrease in self-reported loneliness (Pittman & Reich, 2016).
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Figure 2:
[aesop_video src=”youtube” id=”504U–X1Zk8″ width=”100%” align=”left” disable_for_mobile=”on” poster_frame=”https://cmdstudio.nl/mddd/wp-content/uploads/sites/58/2021/03/Schermafbeelding-2021-03-29-om-19.59.45.png” loop=”on” controls=”on” mute=”off” autoplay=”off” viewstart=”off” viewend=”off” show_subtitles=”off” revealfx=”off” overlay_revealfx=”off”]
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Figure 3:
[aesop_video src=”youtube” id=”FA0qeElMV58″ width=”100%” align=”left” disable_for_mobile=”on” loop=”on” controls=”on” mute=”off” autoplay=”off” viewstart=”off” viewend=”off” show_subtitles=”off” revealfx=”off” overlay_revealfx=”off”]
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Figure 4:
[aesop_video src=”youtube” id=”A5pgM1CUO_E” width=”100%” align=”left” disable_for_mobile=”on” loop=”on” controls=”on” mute=”off” autoplay=”off” viewstart=”off” viewend=”off” show_subtitles=”off” revealfx=”off” overlay_revealfx=”off”]
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Results
- This test showed that the chatbot with the blue colorscheme actually made the respondents feel more stressed compared to the chatbot with the pink colorscheme. One of the respondents said: “I feel more like someone is watching the conversation with the blue chatbot, so I find it more difficult to fill it out.”;
- The chatbot with the GIFs was experienced as the most fun because the respondents felt better understood because of the GIFs and because some GIFs made it easier for them to empathize with the question;
- At the same time, some GIFs were perceived as more confrontational due to their intensity. All respondent said they would prefer the chatbot with the GIF’s if they were more personalized.
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Conclusion and Reflection
In de next iteration I need to use a playful colorscheme. Further the next chatbot needs to be image-based. However, the chatbot needs to become more personalized.
[aesop_parallax height=”30px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Iteration 3
In the next iteration I wanted to make a chatbot that is more personalized. I wanted to make a chatbot that responds based on someone personality type. Due to a time-limit I decided to only work out only one personality types using the enneagram type 2w3. However, this theory isn’t scientifically tested, it is valuable in case of the goal of this research because the enneagram theory also takes different levels of mental health in consideration. Further, I wanted incorporate more gamification into this chatbot that responds to the users need to get a reward based on their actions.
[aesop_parallax height=”10px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
First I worked out a personality for the chatbot and a conversation (figure 5). This personality will contributed to a more personalized experience for the user. Further, I designed a short ‘puzzle’. The users will receive a piece of the puzzle while interacting with the chatbot as a reward.
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Figure 5:
[aesop_image img=”https://cmdstudio.nl/mddd/wp-content/uploads/sites/58/2021/03/Sam_chatbot.png” panorama=”off” imgwidth=”60%” offset=”0x” align=”left” lightbox=”off” captionsrc=”custom” captionposition=”left” revealfx=”off” overlay_revealfx=”off”]
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Next I started creating the chatbot in IBM Watson because I wanted the chatbot to learn to interact with the user based on their input. However, I quickly realized that IBM Watson didn’t have a the functionalities I needed. That why I switched to Power Virtual Agents from Microsoft. With this program I was also able to let the chatbot take initiative in the conversation (figure 6).
[aesop_parallax height=”20px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Figure 6:
[aesop_video src=”youtube” id=”dTYU_A38dEc” width=”100%” align=”left” disable_for_mobile=”on” loop=”on” controls=”on” mute=”off” autoplay=”off” viewstart=”off” viewend=”off” show_subtitles=”off” revealfx=”off” overlay_revealfx=”off”]
[aesop_parallax height=”50px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Conclusion
The main thing I’ve learned by creating these chatbots is how to create a more advanced chatbot without any coding knowledge. Second, I have created knowledge of how a chatbot can be a tool for a young adult to feel more comfortable expressing their feelings. However, if a chatbot really wants to influence the stigma that is present on the topic of loneliness, it must adapt better. This must be done by better training the AI, for example to be able to discover someone’s personality type and to be able to give the user a response based on this discovered personality type. In my opinion, the chatbot is best made by actually coding a chatbot instead of using an already existing tool. This can give the chatbot more options, for example to add gamification features.
[aesop_parallax height=”50px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
References:
Abd-alrazaq, A. A., Alajlani, M., Alalwan, A. A., Bewick, B. M., Gardner, P., & Househ, M. (2019). An overview of the features of chatbots in mental health: A scoping review. International Journal of Medical Informatics, 132, 103978. https://doi.org/10.1016/j.ijmedinf.2019.103978
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Killgore, W.D.S., Cloonan, S.A., Taylor, E.C., Dailey, N.S., 2020. Loneliness: a signature mental health concern in the era of COVID-19. Psychiatry Res. https://doi.org/ 10.1016/j.psychres.2020.113117. Epub ahead of print.
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Lucas, G. M., Gratch, J., King, A., and Morency, L.-P. (2014). It’s only a computer: virtual humans increase willingness to disclose. Comput. Hum. Behav. 37, 94–100. doi: 10.1016/J.CHB.2014.04.043
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
McQuaid, R. J., Cox, S. M. L., Ogunlana, A., & Jaworska, N. (2021). The burden of loneliness: Implications of the social determinants of health during COVID-19. Psychiatry Research, 296, 113648. https://doi.org/10.1016/j.psychres.2020.113648
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Myers, I. B. (1962). The Myers-Briggs Type Indicator: Manual (1962). Consulting Psychologists Press.
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Pittman, M., & Reich, B. (2016). Social media and loneliness: Why an Instagram picture may be worth more than a thousand Twitter words. Computers in Human Behavior, 62, 155–167. https://doi.org/10.1016/j.chb.2016.03.084
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Qualter, P., Vanhalst, J., Harris, R., Van Roekel, E., Lodd, G., Bangee, M., … Verhargen, M.
(2015). Loneliness across the lifespan. Persp. on Psych. Science, 10, 250–264.
[aesop_parallax height=”5px” parallaxbg=”on” captionposition=”bottom-left” lightbox=”off” floater=”off” floaterposition=”left” floaterdirection=”none” overlay_revealfx=”off”]
Russell, D., Peplau, L. A., & Cutrona, C. E. (1980b). The revised UCLA Loneliness Scale: Concurrent and discriminant validity evidence. Journal of Personality and Social Psychology, 39(3), 472–480. https://doi.org/10.1037/0022-3514.39.3.472