Algorithmic Mirror 


Helping adolescents see and shape their algorithmic selves

Role  First author
Year  2024- (Ongoing)
Link https://www.media.mit.edu/projects/algorithmic-mirror/overview/
Key words  algorithmic awareness, adolescents, social media
Algorithmic Mirror is an interactive visualization tool that simulates the hidden profiling practices of social media platforms. Developed in collaboration with MIT Media Lab’s Viral Communications, Oxford's Synthetic Society Lab and Oxford’s Child-Centred AI Lab, the tool experiments with large language model–based sentence embeddings to reimagine how online activity might be categorized and interpreted.

Rather than offering dashboards of quantitative metrics or metadata counts, Algorithmic Mirror generates an interactive, explorable landscape: a speculative “profile” of young people’s digital traces. The project does not uncover actual platform models but instead constructs possible mirrors of algorithmic judgment—artificial reflections that expose how identity could be abstracted and rearranged through data.

By turning opaque inferences into interactive spatial metaphors, the tool invites reflection on how children’s online footprints may be silently transformed into profiles, and how inhabiting such mirrors might alter one’s sense of self.






Temporal evolution of digital selves
Users saw the temporal evolution of their algorithmic selves.

“When YouTube tries to understand what I like, my question is how would it try to track my interest over time and project new interests, or would it just like take me as I currently am and give me exactly what I like?” - P9

YouTube self
TikTok self
Netflix self


User could see their YouTube self, TikTok self, and Netflix self in a unified explorable landscape.
“It made me reconsider how my interests are distributed across platforms. It's as if different personalities exist within each platform”  P2



Credits


Yui Kondo, Oxford Internet Institute
Kevin Duneell, MIT Media Lab
Isobel Voysey, CS@Oxford
Qing Hu, HCII@CMU
Victoria Paesano, CS@MIT
Phi Nguyen, CS@CMU
Dr. Jun Zhao @Oxford Child-centred AI Lab
Dr. Luc Rocher @Oxford Synthetic Society