It's a bit of a tragedy to be honest. We/I lived through both developing social networking and big data, and it's pretty clear that the community made huge mistakes that both hurt loads of people and we can't really take back or clear up at this point. Bad things happened to innocent people. Bad people got power over good people and did bad things with it.
AI has at least as much potential to be dangerous, ethical constraints on it are important when you consider the failure of common sense and community good will for SN and BD. However, every berk going latched onto the idea that they could write some sort of rubbish about "applying type 1 and type 2 thinking in AI design" or "why robots can't really be people" and have founded community dedicated to promoting themselves and getting money at the expense of insight and real work and real constraints developed by people who know what they are on about and can really help.
I think most of the "AI ethics" thing is two things: first, for people that are actual AI practitioners, it makes "AI" seem a lot more powerful and interesting than it is. I think AI is a small part of the example problems you mentioned. Second, it's a way for humanities types to wedge themselves into a hot, high paying field.
I think it's unwise to minimize the impact and significance of AI at this time - this is the same sort of thinking that prevented proactive and effective management of Social Networking and Big Data technologies. We have to take it seriously this time round, and if it proves to be a y2k thing - well no problem.
You are right about the humanities types - the otherside of the coin is that there is a definite class of insight that should come out of social sciences and history that is relevant here. Unfortunately those fields are dominated by bullshit artists which means that the signal to noise ratio is very very low.
AI has at least as much potential to be dangerous, ethical constraints on it are important when you consider the failure of common sense and community good will for SN and BD. However, every berk going latched onto the idea that they could write some sort of rubbish about "applying type 1 and type 2 thinking in AI design" or "why robots can't really be people" and have founded community dedicated to promoting themselves and getting money at the expense of insight and real work and real constraints developed by people who know what they are on about and can really help.