Haters will say it’s AI: Mapping AI-generated brain rot content communities on TikTok & YouTube

—

da

MSMC admin
in

Team Members

Amina Djedidi, Çiğdem Erd, Hanaa Hamidi, Sinin Nakhle, Anastasiia Osipova, Victoria Ro, Camilla Volpe (Project Leader)

1. Introduction to the project

This project explores how the online phenomenon commonly referred to as brain rot relates to the emergence of what we call synthetic publics — collectives that aggregate around AI-generated or AI-influenced content on social media.

Brain rot is defined as the supposed deterioration of a person’s mental or intellectual state resulting from the overconsumption of trivial or unchallenging online content (Owens, 2025; Oxford Dictionary). It has also been conceptualised as a participatory genre through which particularly young users intentionally decompress and resist productivity pressures (Owens, 2025). Such brain rot videos are typically described as childish and unserious, offering no educational benefit and being deliberately non-productive.

The participatory framing proposed by Owens (2025) suggests that young people — particularly members of Gen Z and Gen Alpha — consciously choose brain rot content as a way to unwind and detach from the pressures of “useful” media consumption. Studies involving teenagers on TikTok indicate that those born after 1995 are especially susceptible to “doomscrolling” and engagement with low-value content (Yousef et al., 2025).

By design, brain rot videos fit the affordances of short-form platforms: they exploit brevity, quick cuts, unexpected visuals, loud audio cues, and surreal humor to capture fleeting attention spans (Yuanyuan et al., 2025). Although specific academic research on brain rot style is still limited, broader studies on short-form video demonstrate that such clips are optimized to grab attention through sensory overload and pacing (Jiang & Ma, 2024; Violot et al., 2024).

Against this background, our project aims to map “synthetic publics”, understood as collectives that form around synthetic or AI-generated content. Through a digital methods approach, we investigate how users, algorithms, and synthetic media co-produce new modes of engagement that blend humor, automation, and attention economies. By tracing the circulation of brain rot and AI-generated content, we aim to understand how participatory practices and synthetic content together shape emerging forms of online publicness.

2. Research Question

How do publics on TikTok and YouTube interact with AI-generated brainrot content, and how does presentation style of these videos mediate this engagement?

3. Methodology

Platform-Specific Sampling Design

Our analysis focused on two major video platforms: TikTok and YouTube. For each platform, we collected 9–12 videos tagged with brain rot-related hashtags, focusing on recent content produced within the last six months. To minimize algorithmic bias and prevent personalization effects, we created dedicated research accounts using Firefox’s incognito mode, following best practices in digital methods for avoiding “contamination” and “identification” by the platforms’ recommendation systems.

Hashtag-Driven Content Discovery

Content discovery followed a hashtag-based sampling strategy. We began with the core tag #brainrot, which was expanded through AI-assisted hashtag snowballing to identify related and emergent tags.

  • TikTok hashtags: #aibrainrot, #aislop, #aigeneratedbrainrot, #aibrainrotcore, #genaibrainrot
  • YouTube hashtags: #aibrainrot, #aigeneratedbrainrot, #genaibrainrot, #aibrainrotvideo, #aibrainrotcompilation, #aibrainrotshorts

This approach enabled us to capture both the aesthetic and discursive boundaries of the AI brain rot phenomenon across platforms.

Data Collection

We combined web scraping, metadata analysis, and qualitative coding.

Technical Tools:

Multi-Modal Content Analysis

We collected engagement indicators such as views, likes, shares, and comment counts to assess patterns of visibility and interaction.
Later, we conducted comment analysis using NVivo 13, supplemented by Python-based emotion detection scripts to identify affective tones and reaction types in user responses.
A codebook was developed iteratively to categorize sentiment expressions (e.g., humor, irony, disgust, admiration) and to ensure consistency across the research team.
This multi-modal strategy allowed us to trace how synthetic publics engage with brain rot content — not only through algorithmic visibility but also through the emotional and participatory dynamics emerging in comment sections.

4. Findings

3.1 Engagement Metrics

Descriptive statistics indicate that YouTube videos reached significantly higher visibility than TikTok posts:
YouTube median views: ~1.2M
TikTok median plays: ~132K

However, TikTok posts appear to trigger more direct interaction relative to exposure, as seen through a higher ratio of comments / plays and shares / plays. This suggests that engagement density is stronger on TikTok, while reach is greater on YouTube.

3.2 Sentiment Distribution

Across both platforms, the majority of comments fell into positive + humorous/ironic categories, while hostile or strongly negative sentiment was less frequent.
Neutral and informational comments were less common, signalling a reaction-driven participation pattern aligned with short-form entertainment.

3.3 Linguistic Patterns

Word frequency analysis reveals that top recurring tokens differ per platform:
TikTok → emoji clusters and expressive markers dominate
YouTube → lexical slang terms (e.g., bro, wtf, nah) are most frequent

This illustrates that affective responses and informal language are core communicative elements of AI brainrot publics, but shaped by platform affordances.

.4 Codebook Application Outcomes

Applying the codebook showed that:

  • Humorous / ironic endorsement was the most prevalent sub-category
  • Meme-based comments and off-topic/joke participation were common
  • Some comments displayed mixed sentiment, combining admiration and disgust

These trends confirm that emotional ambiguity is central to how brain rot content is processed and expressed by audiences.

5. Discussion

5.1 Different Platform Behaviors

Our comparative analysis revealed clear platform-specific behaviors in how users interact with AI brain rot content. On TikTok, comment sections are characterized by an intensive use of emojis, which function as affective shorthand to express humor, irony, or shock. By contrast, YouTube users tend to rely more on text-based slang (e.g., “bro,” “wtf,” or “nah this can’t be real”), signalling a more discursive mode of engagement. These linguistic preferences suggest that platform design and affordances shape community language: TikTok’s fast-paced, mobile-first interface privileges visual and emotive expression, while YouTube’s comment threads enable longer, text-based interactions. Interestingly, TikTok users often tag friends and creators, even under seemingly “passive” brain rot clips. This tagging practice may constitute a form of hidden community-building through individual consumption, where algorithmically recommended content becomes a shared social reference.

Overall, each platform fosters distinct interactional norms and lexical repertoires, supporting the idea that brain rot communities are not uniform but are platform-contingent formations emerging from specific engagement architectures.

5.2 Cross-Cultural Community Formation

Comment sections across both platforms display a notable multilingual presence, with frequent appearances of Arabic, Turkish, and Thai alongside English and Italian.
Users often experiment with phonetic transliteration to communicate across linguistic boundaries, demonstrating active cross-cultural negotiation within brain rot publics.

Italian brain rot content, in particular, achieved global visibility — suggesting that absurdity, humor, and AI-generated surrealism are perceived as internationally relatable affective cues, even when detached from local linguistic contexts. This raises the question: how do globally circulating AI-mediated absurdities become locally meaningful?

5.3 Meta-Awareness and Cultural Integration

A subset of users explicitly referenced “Dead Internet Theory”, a conspiracy-adjacent discourse suggesting that much of online content is AI-generated or automated. Such references indicate a degree of meta-awareness among users: they recognize that they might be participating in synthetic ecosystems, where content, users, and affective reactions are co-shaped by algorithmic processes.

This reflexivity demonstrates that brain rot publics are not merely passive consumers but culturally literate participants, aware of the blurred boundaries between human and synthetic media. Through humor and irony, they integrate this awareness into their participatory practices, effectively embodying the dynamics of synthetic publics the project aims to map.

6. Conclusion

This project shows that AI-generated brain rot content fosters distinct yet interconnected publics across TikTok and YouTube. While engagement patterns and linguistic practices differ by platform, users collectively form affective, playful, and sometimes self-aware communities around synthetic media. These findings highlight how platform design and AI-generated aesthetics shape emerging modes of participation within synthetic publics.

This study is limited by the relatively small TikTok sample size and by the exclusion of key expressive elements such as emojis, abbreviations, and video aesthetics, which are central to brain rot communication and community-building. On YouTube, technical constraints of scraping tools restricted our access to Shorts, limiting cross-platform comparability. Future research should expand data collection, include systematic visual analysis, and employ more advanced scraping and sentiment-detection methods.

7. References

Jiang, Q. & Ma, L. (2024). Swiping more, thinking less: Using TikTok hinders analytic thinking. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 18(3). https://doi.org/10.5817/CP2024-3-1

Owens, E. (2025). ‘It speaks to me in brain rot’: Theorising ‘brain rot’ as a genre of participation among teenagers. New Media & Society. https://doi.org/10.1177/14614448251351527

Oxford Word of the Year (2025). Retrieved from: https://corp.oup.com/word-of-the-year/

Park, J. & Oh, C. & Kim, H. Y. (2024). AI vs. human-generated content and accounts on Instagram: User preferences, evaluations, and ethical considerations. Technology in Society, Elsevier, vol. 79(C). https://ideas.repec.org/a/eee/teinso/v79y2024ics0160791x24002537.html

Renzella, J. & Rozova, V. (2024). The ‘dead internet theory’ makes eerie claims about an AI-run web. The truth is more sinister, https://www.unsw.edu.au/newsroom/news/2024/05/-the-dead-internet-theory-makes-eerie-claims-about-an-ai-run-web-the-truth-is-more-sinister

Violot, C., Elmas, T., Bilogrevic, I. & Humbert, M. (2024). Shorts vs. Regular Videos on YouTube: A Comparative Analysis of User Engagement and Content Creation Trends. In ACM Web Science Conference (WEBSCI ’24), May 21–24, 2024, Stuttgart, Germany. ACM, New York, NY, USA 11 Pages. https://doi.org/10.1145/3614419.3644023

Yuanyuan, G., Ying, H., Jinlian, W., Chang, L., Hohjin, I., Weipeng, J., Wenwei, Z., Wei, G., Guang, Z., Qiong, Y., Pinchun, W., Manman, Z., Xin, N., Qinghua, H. & Qiang, W. (2025). Neuroanatomical and functional substrates of the short video addiction and its association with brain transcriptomic and cellular architecture. NeuroImage, Volume 307. https://doi.org/10.1016/j.neuroimage.2025.121029

Yousef, A. M. F., Alshamy, A., Tlili, A., & Metwally, A. H. S. (2025). Demystifying the New Dilemma of Brain Rot in the Digital Era: A Review. Brain Sciences, 15(3), 283. https://doi.org/10.3390/brainsci15030283