Research explores how TikTok’s recommendation system may influence teen beliefs

A new research study on the TikTok algorithm is about to investigate how the “For You” page might affect the worldview, mental health, and behaviour of teenagers online. The study conducted by Georgia Tech gathered data from over 10,000 teenagers in the UK who were willing to share their archives of TikTok data for free under General Data Protection Regulation (GDPR) regulations.

Instead of looking at posts, the study looks at view history to see how the recommendation algorithm on TikTok is affecting what the teenagers are viewing and absorbing.

The project is led by Georgia Tech Professor Munmun De Choudhury in collaboration with the University of Cambridge, Researcher Amy Orben, and UCLA Researcher Homa Hosseinmardi. Unlike earlier social media research that focused on what teens publicly share, this study looks at passive consumption, including late-night scrolling sessions that leave no public trace.

TikTok’s algorithm

The "For You" page on TikTok uses machine learning to predict what users will watch next. According to researchers, the predictive algorithm is not an impartial one. It may inadvertently influence interests, support beliefs, and even promote themes, ranging from body image trends to politics.

Harmful content has been a concern for some time. A 2022 report by the Center for Countering Digital Hate found that test accounts registered as 13-year-olds were rapidly shown eating disorder and self-harm-related videos. TikTok has previously disputed such findings, arguing that experiments do not reflect real user behaviour.

The Georgia Tech team plans to use AI-generated simulations to map potential “rabbit hole” pathways within TikTok’s algorithm. By recreating how small signals may alter recommendations, the researchers hope to identify patterns that could affect teen mental health and digital wellbeing.