Thanks Carlos, I completely agree that a machine doesn’t need to be able to explain itself to be useful.
I would however disagree that we only care if they do a good or poor job. In some of the most exciting use cases such as autonomous driving or medical diagnosis, society today would demand an explanation if something went wrong, particularly if this resulted in loss of life. The relatively recent Tesla crash is a clear example here.
I agree that the potential of these technologies is already being demonstrated today, but I believe that this issue is still a hurdle on the path to mass adoption.