Innovative Modeling of Deep Fake Dynamics on the Population by using Artificial Neural Network

This paper presents an innovative model for understanding the dynamics of information propagation within social networks. Incorporating cognitive biases, follower influence, and temporal decay, we propose a mathematical framework to simulate how information spreads through a network of individuals connected by varying degrees of trust, familiarity, and social influence modeled as a Neural Network. Our model accounts for the role of confirmation bias, the bandwagon effect, and fact-checking delays to capture real-world phenomena that affect the spread of true and false information alike. This innovative model is based on a hybrid approach that uses components based both on static Neural Graphs (to capture the structure of the social network) and on models of epidemic diffusion of information (to model the dynamics of propagation over time). To test the model we used LLMs and open source data to generate opinions respect to different topics in the population network based on different factors, such as age, gender, social status, educational level etc. The authors introduce in the network different messages (real and fake) trough an embedding layer in order to understand the spread of information. The proposed model is validated against state-of-the-art approaches and aims to enhance predictive accuracy in fields such as misinformation control and viral marketing.