From idea to reality. The role of artificial intelligence in creating manipulative messages (on the example of Claude and Gemini services)

Abstract

The article is devoted to the peculiarities of the current use of modern generative language models for creating manipulative messages. The authors explored how neural network algorithms can be applied in contemporary public discourse, which is undergoing rapid transformation due to advancements in deep machine learning technologies. An analytical review of the capabilities of generative language models, such as Claude and Gemini, is presented, focusing on their potential to generate texts employing suggestive and semantic manipulation techniques. The study includes an authorial experiment aimed at examining the practical implementation of manipulative techniques in generated texts, as well as their reverse analysis. The authors identified key manipulation mechanisms used by these models and proposed recommendations for their ethical application. The article highlights the importance of a multidisciplinary approach to regulating and developing standards for generative language models to minimize public risks and maximize societal benefits. Particular emphasis is placed on the necessity of enhancing algorithmic transparency, preventing social biases in data, and introducing feedback mechanisms. Additionally, the paper stresses the critical role of educating users to critically evaluate information and fostering media literacy, which emerges as a key factor in safeguarding against manipulative practices in the context of society’s digital transformation.