Using Large Language Models for Simulating Social Networks

Online platforms, especially social media, can present a set of challenges and harms to users such as toxicity, misinformation, and bullying. To mitigate this, social media companies implemented a number of intervention and moderation policies. However, research highlights the inefficiency, and sometimes the counter effects of these interventions. This PhD thesis work tackles this issues by proposing a different approach to designing intervention policies by simulating social media using Large Language Models.