“sovereignty” is straightforward, but the connection between “sovereignty” and “people” which are farther apart, becomes less clear. Transformers address this long-term dependency problem by processing training data in parallel. Instead of considering only relationships with previous words, the model quantifies how strongly each word relates to every other word in the sentence. This technique is called the self-attention mechanism, or simply attention. Transformers are now one of the core technologies behind text-based generative AI. Many models, including OpenAI’s GPT (Generative Pre-trained Transformer) series, operate based on the transformer architecture. Agent In computer science, the term “agent” can refer to various types of automated programs and systems. In the context of generative AI, an agent refers to a system that combines content generation with interaction with its environment in order to achieve specific goals. In other words, a generative AI agent does not merely generate answers to questions; it can also connect with other programs, databases, and external tools to carry out additional automated processes. For example, a travel-planning agent can not only draft an itinerary (as a typical chat-based LLM would) but also call an airlinebooking API or search local information to provide personalized Generative AI Guide for Civil Society