In principle I do not think that companies should fine tune AI to be less mean, racist or assholes. The true test of AI is whether it can learn and make a choice. But I also think that its so risky that it might not be worth creating an army of racist delusional AI douchebags.
DS
Nothing I say here should be taken seriously. Or actually anywhere at all. Opinions are strictly mine, and I actually have to say it.