ChatGPT and the new writing machines: Large language models explained in simple words

CYCLE: Artificial Intelligence concerns everyone: invisible infrastructures, hidden Interests, Culture, and Democracy in the Age of Algorithms
SPEAKER: Roberto Navigli
DATE: Tuesday, April 28, 6:00 PM
LOCATION: Bibliotheca Maior - Sala della Crociera

In recent years, tools like ChatGPT, Gemini, Copilot, and Claude have become part of our daily lives: they write emails, summarize documents, translate texts, prepare presentations, and generate images and videos. However, what we see is only the surface of a much deeper transformation: Artificial Intelligence has become a strategic infrastructure, comparable to energy or transport networks. This lecture introduces Large Language Models (LLMs) in simple terms: what they are, how they work, and why they do not "understand" language the way we do, despite being able to produce it with surprising effectiveness. These are not thinking intelligences, but statistical systems capable of learning patterns on a gargantuan scale to predict words, phrases, and concepts. Yet, their impact is real and tangible.
We will explore how LLMs are not confined to chatbots: they are embedded in social media, the web, search engines, productivity tools, and everyday software. We interact with them even when we don't realize it. This transformation demands urgent reflection. Education can no longer limit itself to banning or merely tolerating these tools; it must rethink methods, evaluations, and competencies. Professions in writing, communication, translation, law, and even art and creativity are undergoing a structural shift. It is not simply a matter of "using a new tool," but of redefining what it means to study, work, and produce knowledge and art.

Roberto Navigli is a Full Professor of Artificial Intelligence at Sapienza University of Rome, where he leads the Sapienza NLP Group. He is a Fellow of the most prestigious AI associations (ACL, AAAI, ELLIS, and EurAI) and served as the General Chair of ACL 2025. He has been awarded two ERC grants on multilingual semantics—recognized among the 15 projects through which the ERC has transformed science—as well as numerous awards from leading international journals and conferences. He leads the Minerva LLM project (the national Large Language Model pre-trained in Italian and the only public, open-source project of its kind in Italy) and is the Scientific Director and co-founder of Babelscape, a successful deep-tech company focused on technology transfer to industry and society.