Social Media Machine reveals to young people how social media algorithms collect their data

Developed in the Generation AI research project, Social Media Machine (Somekone) exposes social media’s hidden mechanisms to schoolchildren, showing how scrolling, liking and other forms of engagement build a user profile.

Most of us use social media daily, yet few of us know how its algorithms actually function.

“Platform users are often unaware of how much user data are being collected, what kind, and for what purpose. They don’t understand the core mechanisms behind data collection and use,” says Professor Teemu Roos, whose research interests include AI.

This may breed misconceptions: informal and often inaccurate assumptions about how social media algorithms and platforms operate.

It also presents a significant challenge for teaching media literacy in schools.

“Teaching in schools focuses on information security and misinformation, while the mechanisms underpinning social media often go unexplored. To help pupils understand how algorithms really work – how different social media bubbles form, how emotions are amplified and how behaviour is shaped – you need to dig beneath the surface.”

Until now, though, the necessary tools simply have not existed.

Researchers developed Social Media Machine for exactly this reason: to help pupils see what is under the hood.

Every click leaves a trace

Users scroll through a stream of images in a closed environment similar to Instagram, behaving just as they would on any other social media site: liking, following, sharing, commenting and reacting with emojis.

“All the while, the machine is collecting data on all of this, forming a visual picture of the user data gathered and the user profile created, and of how the platform ends up recommending more content.”

Social Media Machine tracks everything: how long images are viewed, what earns likes, what gets commented on and how long those comments run, how users react, and what they follow and share. Pupils can use a second device at the same time to monitor data on their own or their friends’ scrolling.

This collected activity builds a profile reflecting the pupil’s interests, inferred from their scrolling behaviour and displayed as a word cloud, for instance. Drawing on these profiles, Social Media Machine also recommends further images, explaining to the pupil why a particular piece of content has been chosen, perhaps because it resembles something they have previously liked or because other, similar users have liked it.

“Learners can see for themselves how even the smallest actions, such as pausing for a moment to look at something, are stored as part of their digital footprint.” 

Visualising and explaining data verbally makes what is being collected immediately more concrete: every action a user takes shows up as a change in their browsing history and profile.

“Everyone’s profile in the class is shown on the teacher’s screen as a real-time network, with pupils who share similar interests grouped close together. This gives pupils a tangible way to see how one learner’s actions shape the recommendations that others receive.”

Learners’ data protection has been central to the tool’s design: during lessons, all data are stored locally on the teacher’s computer and automatically deleted once the lesson ends. Social Media Machine does not transmit identifiable data to external servers.

The tool has already been trialled with roughly 400 children across dozens of Finnish schools.

“Social Media Machine supports experiential, hands-on learning. 

“I’d love to see Social Media Machine used widely in schools, and at home too, for that matter. Learning through an app, a format most young people are already familiar with, rather than through a traditional lesson or textbook, brings out just how relevant the subject is to everyday life, and encourages learners to join in conversations about social media and express their own opinions. This would be an important way to make sure that discussions about restricting social media use don’t take place solely among adults, leaving young people out of the conversation.”

The project has also received a grant from the Jenny and Antti Wihuri Foundation.