UNDERSTANDING GENERATIVE ARTIFICIAL INTELLIGENCE FROM THE LANGUAGE PERSPECTIVE

Abstract

This article explores text generation through Generative Artificial Intelligence (Gen AI), focusing on the interactions between AI and Language. The research aims to elucidate the underlying mechanisms of text-to-text generation tools, such as ChatGPT, placing it within the broad field of AI, an interdisciplinary area that counts Linguistics as one of its pillars. The investigative journey spans from the theoretical foundations of AI, traversing fields like Natural Language Processing, Machine Learning, Deep Learning, and the revolutionary Transformer Architecture. The study emphasizes Distributed Semantics as the enabler for the emergence of Large Language Models. Highlighting the confluence of technology with natural language, the paper outlines the capabilities, limitations, and challenges of these models, especially in generating textual content from input commands, known as "prompts". In this context, our primary goal is to understand the fundamentals and workings of Generative AI and its intersections with Language, enabling the use of tools geared towards text-to-text generation, like ChatGPT.