Given that we do not know what in intelligent wetware results in the emergence of sentience, it may occur without any intent on the designers’ parts to create it.
Given that we do not know how to recognize sentience (other than the assumption that those built like us and responding like us have it), there is no reason to have confidence that we’d recognize or accept it if it was in front of us, especially if it didn’t look or respond like we do.
Networked AI systems could be sentient now and we’d never know.
As to the OP- one approach is QALY which looks at quality adjusted life years. Generally the the US we consider medical interventions reasonable if the cost about $50 to 60K per QALY. Not consistently applied of course.