Yes, this could be only posible when been programmed with theories and philosophies propounded by other researchers and questions been diverted to the mails of expects to answer.be

There's a team in our department that work on affective computing and machine emotions. You can learn about the rsearch here: https://www.researchgate.net/project/NeuCogAr

So, answering your question, eventually they will, but they would probably teach some other morality than what we know. Human morality evolved throughout the time as well. E.g. it was totally ok to burn people in fire if convicted in heresy or witchcraft in middle ages in Europe, many ancient cultures practices human sacrifices to their lords. Hopefully, that time will never come back, but what morality the sentient robots will teach we can hardly predict as of now.

Of course, an AI trained to teach human morality is totally imaginable. On the other hand the strongest tool in pedagogy is a personal example, that means in order for it to work the robot should be a person, otherwise that AI would be nothing but another form of eLearning.