From David Hodgson, London, UK
I fear that the risk of labelling AIs as sentient is even more serious than described. If you were to follow a certain line of argument, terminating something that may develop into a sentient entity could be viewed as murder.
That would mean that all future attempts to make LaMDA-like AIs must be treasured forever once created. Google’s servers – along with those belonging to IBM, Microsoft, Amazon, Nvidia and several universities – would collapse under the load of these misbegotten but eternal creations.
