Do we really need the concept of consciousness in AI research?
As large language models become more fluent, useful, and socially embedded, debates about artificial consciousness have become increasingly visible. Are advanced AI systems conscious? Do they genuinely understand what they produce? And what exactly do such claims mean?
For ai-phi #38, we are delighted to welcome Thierry Poibeau for a seminar.
In this session, Thierry Poibeau will revisit notions such as consciousness, meaning, and understanding, whose definitions often remain vague or unstable in the context of artificial intelligence. Drawing on both the current capabilities and limitations of large language models, especially their lack of perceptual grounding and their tendency to hallucinate, the talk will invite us to question how these concepts are currently mobilized in debates about AI, and what practical consequences this may have for oversight and regulation.
Thierry Poibeau has also recently published "Understanding Conversational AI: Philosophy, Ethics, and Social Impact of Large Language Models" with Ubiquity Press. The book is fully open access.
As always, ai-phi aims to create a space for careful, interdisciplinary reflections at the intersection of AI and philosophy.
📅 Thursday 21 May 2026, 19:00-20:30
📍 Sony CSL (Paris), 6 rue Amyot, 75005 Paris
🔗 Details and registration: https://lnkd.in/e9asvE6j
If you would like to join the after-session drinks or dinner, please register via Lu.ma or WhatsApp, even on the day of the event, so we can book enough tables.
Many thanks to Sony CSL (Paris) for hosting us.
