ABSTRACT: Ever since OpenAI released ChatGPT, LLMs have been applied to the most diverse domains, from education to medicine. However, what they basically do is to look for patterns in huge amounts of text and use those patterns to guess what the next word in a string is. Having no access to the real world, LLMs are therefore great at mimicry but show important limitations when employed for critical tasks, which may affect human well-being, social justice, and access to digital services. This talk will discuss all the above issues, highlighting the risks but also the huge potential of LLMs. BIO: Sara Tonelli is the head of the Digital Humanities research unit at Fondazione Bruno Kessler since 2013. She holds a PhD in Language Sciences from Università Ca’ Foscari in Venice and is a member of the Center of Computational Social Science and Human Dynamics jointly with the University of Trento.