To be clear, I agree that LaMDA is not sentient. Machine learning is not intelligent. > It doesn't do anything because it's literally "turned off" when not pr

To be clear, I agree that LaMDA is not sentient. Machine learning is not intelligent. > It doesn't do anything because it's literally "turned off" when not prompted or given any inputs. I wondered the same thing. It could be argued that it is sentient during the brief moments that it is coming up with a response. Put another way; let's assume that we are ourselves AI in an artificial universe. Would you be able to tell if the universal computer was turned off for a day? So, a model might not experience sentience in the seconds while you are composing a message, but plausibly could while formulating a response to your input.