On an ordinary day, I casually mentioned to a friend, “I’m thinking of dyeing my hair red lately.” So, I opened TikTok to search for some red hair inspiration. After browsing for a few minutes, I suddenly realised something strange—I only typed “red” into the search bar, not the complete “red hair.” But almost all the content TikTok recommended was about red hair. At that moment, I suddenly realised that there seemed to be an invisible hand behind social media, quietly shaping what I was seeing. TikTok seemed to know what I wanted before I even finished typing. So, are we really choosing the content we want to see? But how does TikTok do this? And more importantly, what does it mean when an algorithm starts shaping what we see, think, and even want?
This story led me to a broader insight: artificial intelligence is not just a technological tool, but a form of governance. This blog argues that TikTok’s algorithmic systems shape user behaviour, influence social norms, and raise important concerns about transparency and power in digital society.
Picture: resource from “The Guardian”
What is AI, automation, and algorithm governance?
Using my personal experience with TikTok as an example, I have noticed that it accurately predicts my preferences. After I like a certain type of video or use a specific audio, the content that appears on my preference page afterwards is almost entirely made up of similar videos or content using the same audio. This effectively demonstrates that much of the content we see on social media is influenced by artificial intelligence and algorithms.
TikTok’s algorithm relies heavily on data collected from users. Every like, comment, scroll, and watch time is recorded and analysed by the platform. This process is also highly automated. Once the data is collected, automated systems continuously analyse user behaviour and make decisions without human intervention. These automated decisions determine which content is recommended, promoted, or hidden, thereby shaping users’ experiences in real time.
Katzenbach and Ulbricht (2019) describe algorithmic governance as the way algorithms create social order by structuring information flows and influencing user behaviour. Traditionally, governance was carried out by governments and institutions. However, in the digital age, platforms are increasingly taking on governance roles. Algorithms are not simply organising information — they are also automatically regulating what users see and how they interact online.
Flew (2021) argues that platforms are becoming new regulators that quietly shape user behaviour and control information flows. In the past, governments determined how information circulated, but today platforms like TikTok influence what users see and who they interact with.
Similarly, Crawford (2021) challenges the idea that artificial intelligence is a neutral technology. She argues that AI is not simply “artificial” or “intelligent”, but a system built on power, data, and resources. Because these systems rely on large amounts of user data controlled by a small number of technology companies, AI may reinforce inequalities and concentrate power in the hands of large tech companies.
In summary, these perspectives suggest that social media platforms like TikTok are not just services for social interaction but governance systems driven by artificial intelligence and automation.
How Does TikTok Govern Users?
TikTok personalises the content on users’ homepages. Unlike traditional social media platforms that rely heavily on who users follow, TikTok uses artificial intelligence and automated recommendation systems to predict what users are most likely to watch next. This recommendation system draws on user data such as likes, shares, watch time, and interactions to deliver more personalised content (Bucher, 2018).
With this understanding, we can better see how TikTok shapes user behaviour.
First, TikTok’s personalised recommendation system subtly influences how long users stay on the platform. While users are browsing, the platform continuously recommends more engaging content. TikTok’s algorithm automatically prioritises content that maximises user engagement. In addition, because TikTok videos are typically short—around 15–20 seconds—users may feel that they are not spending much time on the platform. This illusion, combined with continuous recommendations, leads users to spend more time on the platform without realising it. As Gillespie (2014) points out, platform algorithms are not only used to organise information but also to shape user participation and behaviour.
Second, TikTok’s algorithm also influences creators’ content production. Creators often base their content decisions on likes, engagement, and views, which are themselves influenced by platform algorithms. As a result, creators analyse high-performing content and adjust their future work to align with algorithmic preferences. This means that algorithms indirectly influence what types of content are produced. Duffy and Meisner (2022) note that creators frequently adjust their behaviour in response to algorithmic visibility, suggesting that algorithms not only determine what content is seen but also influence what content is created.
Furthermore, users’ interests may also be subtly shaped by the platform. When users spend more time watching a certain type of video, the algorithm recommends more similar content. Over time, this may gradually influence users to develop an interest in that type of content. This process demonstrates that algorithms not only reflect users’ interests but also actively shape their knowledge, preferences, and behaviours (Bucher, 2018).
These examples may lead to a realisation: TikTok’s algorithm is quietly governing user behaviour.
Platforms do not directly control users. Instead, they influence behaviour through automated recommendations and personalised content. This shows that AI-driven platforms like TikTok are not just entertainment tools, but powerful governance systems that shape people’s behaviour, content creation, and the development of interests.
Picture: resource from “Bulletin of the Atomic Scientists”
A serious question: Do you experience strong body image or appearance anxiety?
At this point, you may wonder what appearance anxiety has to do with platform governance. The following case illustrates how platform algorithms can influence real-world experiences.
In 2022, the Center for Countering Digital Hate created several TikTok accounts to study the platform’s recommendation algorithm. They set the user age between 13 and 17, corresponding to a critical stage of adolescent development. After creating the accounts and watching a small number of videos related to dieting and body shape, TikTok quickly began recommending more extreme content. The study found that teenage accounts were shown appearance- or mental health-related content every 39 seconds, including dieting, calorie restriction, body comparisons, and cosmetic surgery (Center for Countering Digital Hate, 2025). Although these videos appear to be ordinary personal content shared by creators, when they are continuously pushed by algorithms, they gradually form an invisible norm. As users repeatedly encounter such videos, they are guided to focus excessively on their bodies and appearance and gradually accept the idea that being “thin” is the only standard of beauty. This kind of subtle regulation is particularly harmful for adolescents during a critical stage of development.
More importantly, most of this content is not actively searched for by users but rather reinforced by the platform’s recommendation algorithm. TikTok predicts user interests based on watch time, likes, and interactions. Once users spend time engaging with this type of content, the algorithm amplifies it, creating a feedback loop. As users consume more appearance- and body-related videos, they may gradually develop feelings of anxiety.
Content recommended on social media can have a significant psychological impact on teenagers.
Picture: resource from Yale Medicine
This phenomenon clearly demonstrates the influence of algorithmic governance. The platform does not explicitly tell users how they should view their bodies or appearance, yet through recommendation mechanisms, algorithms subtly shape user perceptions. While users believe they are choosing what to watch, the algorithm is guiding their decisions. The report suggests that this influence is particularly significant for teenagers, as they are in a critical stage of identity formation and are more susceptible to external standards. When platforms continuously push idealised body images, extreme dieting methods, and cosmetic surgery content, teenagers may begin to view these ideas as “normal” or even “necessary.” Over time, this may intensify appearance anxiety and potentially lead to eating disorders, psychological stress, and self-doubt.
Therefore, this case highlights how platform algorithms shape user behaviour and social norms through recommendation systems. TikTok’s algorithm is not only deciding what we see, but also quietly influencing how we see ourselves.
Critical Thinking: What governance risks arise when platforms rely too heavily on AI, automation, and algorithms?
1. Lack of Transparency in Governance
AI-driven recommendation systems automatically analyse users’ background data to predict their preferences. After predicting preferences, the system automatically pushes content to users. This series of automated decisions can raise concerns: TikTok does not fully disclose how its recommendation algorithm works, so how can users question or critically reflect on the content they see? As Katzenbach and Ulbricht (2019) point out, algorithmic governance structures information flows in ways that are often invisible to users. Because these recommendation systems operate automatically through artificial intelligence, users may not realise that the content they encounter is shaped by algorithms.
For example, when users watch videos related to body image, the platform may continue recommending similar content. When harmful content is continuously promoted in this way, the lack of transparency becomes even more concerning.
In the case of appearance anxiety, users may not realise that frequently seeing content related to body image or appearance is not entirely due to personal interest, but rather the result of algorithmic recommendations. When users watch or like such videos, the system continues to push similar content. Over time, users are exposed to increasing amounts of videos about dieting, body comparison, or cosmetic surgery. Although users may feel that they are choosing to see this content, algorithms are constantly guiding their attention. In this situation, users’ choices are not entirely free but influenced by platform recommendation mechanisms, which raises concerns about user autonomy and informed consent.
2. Concentration of Platform Power
The growing concentration of power in platforms like TikTok is another important governance concern. These platforms now not only hold vast amounts of user data but also use AI and algorithms to determine what content users see. This means platforms can control information distribution and influence users’ thoughts and behaviours. As a result, a small number of technology companies can exert significant influence over information flows.
By controlling recommendation systems, platforms decide which content is more visible and which content may be overlooked. Unlike traditional media, social media platforms are not just channels for distributing information; they also shape which voices are amplified. When platforms prioritise certain types of content, those topics are more likely to become popular, while alternative viewpoints may be harder for users to encounter. Over time, this mechanism may influence public attention, social trends, and even societal values. This suggests that platforms are not only providing content but also shaping how people understand the world.
Flew (2021) argues that digital platforms are increasingly becoming new “regulators” because they influence user behaviour by controlling information flows. When such power is concentrated in the hands of a few companies, an important question arises: when platforms hold such significant influence, who regulates them? This remains a major challenge in AI and algorithmic governance.
A Moment of Realisation
From the example above, we can see that artificial intelligence and algorithms are no longer just technical tools — they have become a new form of governance. Through automated recommendation systems, platforms quietly influence what we see, what we pay attention to, and even how we perceive ourselves and the world. Unlike traditional governance, this form of governance is not implemented through rules or laws, but through data, algorithms, and automated systems that gradually shape our experiences during everyday use.
This also raises several important questions: When we browse content on social media, are we really making free choices? Or have our choices already been guided by algorithms? When a small number of technology companies hold such immense influence, should platforms take on greater social responsibility? And how should governments and society regulate these technologies?
Picture: resource from Care Clinic
As artificial intelligence continues to develop, these questions are becoming increasingly important. After reading this blog, you might start to wonder the next time you scroll through videos: Is this really what I chose to see? Or is someone — or rather, an algorithm — making those decisions for me?
Be the first to comment