Social media's future is messy, AI models show

New research utilizing AI models simulating user behavior suggests that the inherent architecture of social media platforms structurally embeds dynamics leading to echo chambers and polarization. These effects can emerge even without algorithmic filter bubbles, as users tend to leave communities with a high proportion of dissenting opinions. Surprisingly, filter bubbles, often blamed for homogeneity, could potentially act as a stabilizing force by ensuring users encounter like-minded individuals, thus fostering tolerance for diverse viewpoints. The study analyzed subreddit dynamics, observing that linguistic divergence from community norms correlated with user departure, reinforcing the feedback loop towards homogeneity and extremism.