A Google engineer was suspended after sharing a document suggesting that Google's LaMDA conversation model may be sentient. But if a machine was sentient, how could we tell? What does the Turing Test have to do with it? And can machines think?