Learnscrolling: Harnessing Attention Economy Mechanics for Incidental Learning via Synthetic Social Media
Year: 2026
Type: conference-paper
Abstract: The average person spends over two hours daily on social media, much of it in passive, low-value doomscrolling that users report as unsatisfying yet find difficult to stop. These feeds are engineered for engagement through mechanisms including algorithmic personalization, social proof, variable reward schedules, and frictionless content delivery. This paper explores repurposing such mechanisms for learning within a synthetic social media platform, a feed where every participant is a large language model agent, tuned to a single real user who is the sole genuine node. While such infrastructure could manufacture an unhelpful echo chamber of synthetic consensus, our focus is on its potential to deliver incidental education through what feels like natural social scrolling. We present a conceptual framework and proof-of-concept prototype grounded in established educational principles including social learning theory, spaced repetition, and stealth education. Where existing agent simulations typically study collective dynamics among agents, we focus on a single user as the sole human on a personalized learning platform. The framework and prototype demonstrate how agents with distinct personas can introduce, scaffold, and reinforce learning objectives through varied feed interactions, suggesting a mechanism for incidental learning utilizing the same mechanics that currently drive doomscrolling.
Institution: University of Southern Denmark
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7221242541