ChatGPT and Beyond: A Journey into the World of Generative Artificial Intelligence and Its Potential
Published 2025-02-11
Keywords
- Generative artificial intelligence,
- ChatGPT,
- Large Language Models,
- Retrieval Augmented Generation,
- Agentic AI
Copyright (c) 2025 Pietro Segreto
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Abstract
On November 30, 2022, OpenAI launches ChatGPT. The chatbot records an extremely high number of interactions within a few months, becoming a worldwide success and managing to pervasively spread the concept of generative artificial intelligence in public debate. But what exactly is meant by generative artificial intelligence, and what technological architecture underpins a tool like ChatGPT?
This contribution starts from these questions to explore the potential offered by the use of generative AI models, taking a look at the hottest topics in recent scientific research on this subject. The ultimate goal is to offer a scientifically up-to-date overview that allows one to navigate a new and fluid field like the current one, where technological innovations are piling up at such a high rate that it becomes difficult even for insiders to stay updated.
The first part of the article frames the topic, clarifying the concept of generative artificial intelligence and focusing specifically on Large Language Models and their architecture. Far from wanting to delve into the statistical and engineering specifics that characterize this technology, the intent is to help acquire an intuition about its functioning, in order to reflect in an informed way on the potential and risks of these tools.
Subsequently, the article analyzes the most interesting themes of current scientific research on generative AI. Among these we will discuss about Retrieval Augmented Generation (RAG), which improves text generation through access to updated and contextual information; and agentic AI, which introduces intelligent agents capable of making autonomous decisions.
The contribution closes with a reflection on the significance of generative artificial intelligence for dealing with data today, trying, in light of what was previously exposed, to indicate possible paths of development and integration of this technology that may be of high interest.