Sociocultural and Information Security Issues in the Implementation of Neural Network Technologies in Chat-bots Design

Abstract

Neural networks are trained on actual material. People in their communication use both socio-cultural rules of courtesy and good taste, and take into account threats to protect data, based on long-term experience and long-term risk prediction. Human behavior provides neural networks with examples of both appropriate behavior and manners with limited application depending on the context, as well as unacceptable and unacceptable behaviors, such as rude emotional discharge or information disclosure in associated metadata. Based on the analysis of several cases of the implementation and functioning of chatbots, the main groups of ethical problems and the main tasks in the field of information security are shown, key approaches to ensuring etiquette and data protection are identified, and proposals are formulated for procedures for machine learning in relation to chatbots in corporate ecosystems.