Abstract
My technical capstone project and STS research paper are both centered around recommendation models. Recommendation models are a technological tool utilized in everyday devices and apps to provide users with a more personalized and engaging experience. My capstone team worked with dinemait, a startup company, that targets the personalization benefit of recommendation models for its users. Meanwhile, my research paper explores the engaging aspect of TikTok’s recommendation model. Both pieces are aimed at developing a deeper understanding of the power of recommendation models, with my capstone project focused on seamlessly integrating this tool to promote dinemait’s app and my research focused on the implications of how this technology has integrated into society.
The decision-making process of picking a restaurant to eat at is riddled with a variety of factors such as personal preferences, social dynamics, and an overwhelming number of options. My capstone project addresses this issue by partnering with a startup, dinemait, that utilizes an artificial intelligence (AI) recommendation model to provide curated restaurant suggestions to its mobile app users. This is done by the user first completing the onboarding process where their taste preferences are gathered and then secondly, conversing with dinemait’s chatbot to recieve the recommended restaurants. Our team used a systems-based approach to address the client’s desire to improve the app and increase user engagement by: (1) evaluating the existing app, and (2) improving outreach features and techniques. A study was conducted to gain user-centric data to assess the app’s usability and valuation. We found that there were user experience (UX) issues, inaccuracy in recommendations, and that the app did not always align with a user’s mental model. Next, by researching specific marketing strategies and assessing methods that encourage user interaction with the app, we found that a targeted guerilla marketing strategy and push notifications would benefit reaching a wide audience of diverse users. Ultimately, based on our findings, our team provided these insights to dinemait and recommended the following: streamlining onboarding, clarifying interface terminology, improving feature visibility, diverse marketing techniques based on different platforms and a user’s generation, and connecting with users through A/B tested triggered push notifications. These highlight the importance of having a user-centric approach to engage users to provide them a positive experience.
While a restaurant recommendation app is one beneficial use of recommendation models, my STS research paper investigates another use: social media’s use of recommendation models. My research attempted to understand how the recommendation model of TikTok, specifically, affects polarization by focusing on how it engages its users. The methodology was guided by a literature review that identified similar social media apps with a recommendation system and how they have been affecting society. Because TikTok is a newer app, these findings were extrapolated and applied to determine what similarities TikTok has to these apps (Facebook, YouTube, and Twitter) and then whether these create the necessary conditions for polarization. It was found that TikTok does exemplify traits that can contribute to the polarization of society because it enables cultural lag, cognitive dissonance and confirmation bias, and can be utilized as a mass communication channel. These factors result from TikTok’s recommendation model because they fuel TikTok’s monetization model. TikTok and other social medias typically follow an advertising-based revenue model which means greater user engagement correlates to greater profit. Thus, the recommendation models are structured around this, exploiting the polarization factors. Polarization can occur on different scales. While there have been cases of polarization that lead to violence, users are rather likely to be exposed to the polarization of ideas. Still, polarization creates a division between people, pushing society apart rather than together. This diminishes our ability to work collaboratively toward an improved society by increasing tensions and conflict. Thus, we must employ social media companies to issue preventative measures and encourage users to consume social media responsibly.
By working on both my capstone project and STS research paper simultaneously, I was much more cogniscent of not just the benefits, but also the consequences of utilizing recommendaiton models. When suggesting improvements to dinemait, at the forefront of my mind was to make sure we were not sacrificing the autonomy of users just for the benefit of our client. Thankfully, dinemait users are consciously interacting with the recommendation model technology, so they have more control over their content. In comparison, in TikTok’s environment, a user’s actions still control their content, but users are not always making conscious decisions towards that. For example, a user on dinemait actively converses with its chatbot to prompt-engineer a result they want, constantly having room for feedback. However, on TikTok, how long a user watches a video, what they scroll past, and more are all compiled to determine what content they receive. The latter is less obvious to the user in how they control their outcomes. By working on and researching this duality, I have been prompted to reflect on how other technologies also exhibit obvious benefits, and maybe not as obvious consequences for society. As a society, we should aim to understand the risks of the technology we create, so that it cannot be abused to the detriment of the people. As an individual, we should critically evaluate not only the content we consume, but also all of the technology we use on whether it is benefiting or harming us.
