Is LaMDA Sentient?

David Chalmers (New York University)

Abstract: Artificial intelligence systems – especially large language models, giant neural networks trained to predict text from the internet – have recently shown remarkable conversational abilities.  There has recently been widespread discussion of whether LaMDA, a large language model developed at Google, might be sentient.  Should we take seriously the idea that LaMDA and other large language models are or soon will be sentient?  I will discuss the underlying issue and reasons for and against.