Practical information
17 November 2026
11:30-1pm
Place
ENS, Jaurès, 29 rue d'Ulm, 75005 Paris
LNC2
Abstract:
It is often claimed that AI systems could not be conscious or that we could not tell if they were. In this talk I will argue that we should take the prospect of AI consciousness seriously and that we can make progress in assessing whether given systems are likely to be conscious. I will discuss recent work from Anthropic suggesting that a global workspace is present in LLMs. Then I will discuss whether, if LLMs are conscious, they have experiences akin to pleasure and suffering.
The Cognitive Science Colloquium series is the most attended event of our department, hosting monthly talks by world-renowned experts in various fields of cognitive science, including neuroscience, psychology, linguistics, philosophy and anthropology.
