From TikTok’s “For You Page” to Spotify’s customized playlists, algorithms nowadays drive personalization on almost every major online platform. Such algorithms offer several advantages to users by helping them discover relevant content, reducing choice overload, and increasing engagement. It shapes what we listen to, watch, and shop. However, in reality, does it actually empower us or does it limit us in ways we are unaware of?
The advantages are quite obvious. Algorithms help people discover new and relevant content like movies, videos, songs, and more. On TikTok and Spotify, they give unknown creators and artists a chance to reach many people. Or on Netflix, due to the recommendations, consumers no longer have to scroll endlessly to find a movie they might not have found themselves. For users, this means greater convenience, less choice overload, and more time savings. And for companies, it increases customer engagement, loyalty and revenue.
However, these benefits also come with risks. On a long-term basis it can reduce users’ abilities in preference learning and independent decision-making (Rafieian & Zuo, 2024). In addition, even well-designed algorithms can unintentionally disadvantage specific groups, thereby reinforcing inequalities (Ascarza & Israeli, 2022). For example, in 2019 Apple’s algorithms used to set credit card limits offered women lower limits then men with similar profiles (BBC News, 2019).
Personally, I enjoy and benefit a lot from personalization through algorithms. I frequently discover new items that I would not have found on my own, such as music or clothing. However, I also notice that my choices are strongly influenced by these recommendations. Instead of searching for it myself, I often end up watching or purchasing something based on these suggestions. Therefore, I believe that personalization is most valuable when it is diverse. Occasionally, it doesn’t hurt to show things outside people’s usual preferences, it can create new interests and broaden perspectives. Algorithms should guide people in what they discover, not control it.
References
Ascarza, E., & Israeli, A. (2022). Eliminating unintended bias in personalized policies using bias-eliminating adapted trees (BEAT). Proceedings of the National Academy of Sciences, 119(11). https://doi.org/10.1073/pnas.2115293119
BBC News. (2019, November 11). Apple’s “sexist” credit card investigated by US regulator. https://www.bbc.com/news/business-50365609
Rafieian, O., & Zuo, S. (2024). Personalization, Algorithmic Dependence, and Learning. Cornell SC Johnson College of Business Research Paper (forthcoming). SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5037671
