Recent advancements in artificial intelligence are revolutionising the way we interact with machines. Large Language Models, with well-known exponents such as Llama, GPT, and Claude, are making it possible to engage in fluent, complex conversations with machines. Even more remarkably, such conversations can seamlessly carried out in multiple languages, even switching from one language to another during the same interaction – a capability that is of particular importance in multilingual regions such as South Tyrol. 

Large Language Models are thus transforming the area of AI known under the term of  natural language processing, posing unprecedented research questions to linguists. At the same time, they are already heavily impacting our lives: we use this technology to restyle our emails, to translate administrative documents, to produce essays and stories, to retrieve information, and to answer our curiosities. However, we know little of how this technology works, and it is then difficult for us to understand why it sometimes works flawlessly and other times it miserably fails, and ultimately how to properly use it. To cite only some of the limitations of Large Language Models: they can produce wrong, factually incorrect answers, struggle with reasoning and common sense, interact much better in English than in local dialects and under-represented languages, and often implicitly reproduce human biases. 

The main goal of this panel is to discuss with the general public the current state of the art in natural language processing, the achievements obtained so far, the pros and limitations of contemporary technologies and their possible evolution in the near future, as well their societal and ethical implications and their impact on the job market.

The panel sees the participation of 6 of the most  renowned experts of the topic, reflecting its different facets the view of academia and industry:

• Marco Baroni (Professor at Pompeu Fabra University of Barcelona)
• Pascale Fung (Senior Director of AI Research at Meta; Professor at the Hong Kong University of Science & Technology)
• Roberto Navigli (Professor at Sapienza University of Rome)
• Malvina Nissim (Professor at the University of Groningen)
• Barbara Plank (Professor at the Ludwig Maximilian University of Munich)
• Marco Turchi (Head of machine translation at Zoom Video Communications)

The panel will be coordinated by Silvia Fabbi, a journalist with a degree in cognitive sciences and artificial intelligence. It will interleave moderated discussions with open questions from the public.

The panel will be held in English, with real-time translation in German and Italian.