Our project, ‘Dating Apps: Demographics and the Use of AI,’ examines the range of users on two widely known dating platforms — Tinder and Bumble — with a focus on Artificial Intelligence in online dating. Through an independent experiment, we evaluated the demographic of the first 30 users on our Tinder and Bumble feeds. This was carried out through the same fake profile on both apps, where we generated an AI image that would serve as the first image on our profiles. Both profiles maintained the same settings and preferences, such as being open to all genders and ages between 18 and 80 to increase the range of matches we received. The demographics we sought to compare between Tinder and Bumble users included age, gender, location, race, and what they were looking for on the platforms. Upon collecting the demographics, we experimented with making our own isotypes, using charts and pictographs to convey differences between our datasets visually.
Our inspiration primarily came from the prevalence of fake profiles on dating apps and the theoretical aspects behind catfishing. Before starting our experiment, we looked at similar studies that have been previously conducted by researchers and journalists. These studies gave us an idea of what to expect from users on Tinder and Bumble, and how the algorithm might negatively respond to a fake profile. With Tinder, we expected older, male matches, as opposed to younger, female matches on Bumble. To add a level of sophistication to our experiment and unlike the studies we looked at, we combined the AI image we generated with real images from our personal lives to make our profile appear more realistic. In terms of our website, we were inspired by the aesthetics of Tinder and Bumble to make it look like a dating platform as well. We aimed to reflect similar themes of love and romance across the website and isotypes, using relevant colours and symbols to show our data creatively.
One of the biggest challenges we faced for this project was the ethicality of our fake profile. We were aware of the consequences of catfishing from the start, so we took several measures to limit the problems posed by our fake profile. We made sure not to interact with any users through conversations to prevent any further expectations from the users who matched with us. We also prioritised the privacy of the users, collecting only the demographics we needed and making sure not to share their profiles online. Another issue we faced was the unexpected banning of our Tinder profile, which shortened our study from two days to one. However, we were satisfied with the data we had already collected, as they revealed contrasting demographics between Tinder and Bumble users.
Despite the setbacks we faced, we also had some strengths within our project, particularly the data we collected and the interesting statistics revealed. Unexpectedly, Tinder had a greater percentage of women and younger users overall; Meanwhile, Bumble was dominated by men who were generally older. We believe we have conveyed these demographics creatively and successfully through a range of isotypes and graphs. Our goal of creating a website with similar aesthetics as actual dating platforms has also been achieved through the different pages, popups, and animations throughout the website. Overall, each group member utilised their unique strengths to contribute to the project, whether it be developing visual stories or through technical coding.
With this study being based on the first 30 users in a set location, it would be interesting to see how the demographics may change through different factors, such as evaluating the users across several days, changing our locations, or altering our preferences. The use of AI in this project to create our fake profile also posed some interesting questions, like how a different race or gender may affect the users on our algorithms. If we were to conduct this study again, we would experiment with different characters to see how gender and race play into dating services. We’d also look at other dating platforms, such as Hinge or eHarmony to see if demographics vary across these apps as well. We would also experiment with more AI generation tools and study how the algorithm detects suspicious activity so that we can limit our chances of getting locked out of the profile again.