Pedro Domingos on Machine Learning and the Master Algorithm EconTalk Library of Economics and Liberty

For most of humanity, we've been in the dark about how things "really work", and to a large extent we still are now. That doesn't prevent us from functioning. However, I think Bostrom and Musk are worried about strong AI for other reasons. I do not know whether their concerns are valid or not, but they are not addressed by Domingos when he says that more intelligent machines would just fulfill their original purpose more efficiently. This ignores "Paperclip maximizer" type scenarios: unless both the task and the machine are very limited in their scope, there is potential for the task to interfere with other concerns such as human safety, and this has to be adequately dealt with.