The reason this is a hard question to answer is because it's because it's very difficult to define what it means to be intelligent. IBM's Watson is probably the most famed example of an artificially intelligent "thing". The magic in not knowing exactly what intelligence is almost defines it. Watson seems quite a bit less intelligent when you understand that it plays jeopardy more less by finding the best match to an answer on the internet. Its a huge accomplishment because of how massive the internet is and the difficulty in interpreting English(or whatever language) questions in the first place is no simple task. However, it comes down to a "search, score, and match" operation where Watson probably is not showing any bit of consciousness the way we would normally think of it.

These bits of search and replace are pretty common. You've got intelligence like Siri, Alexa, Google's language understanding(name escapes me) algorithm all trying to understand you and collect data about what you might mean. The auto correct feature that stores spellings of words recently used for auto completion come to mind as simple bits of automation that are not going to go away any time soon.

AI, at least in this form, is going to become more practical and it will put people out of jobs. this is usually what people are talking about when they want to talk about risks. I have a pretty grim outlook personally in that I think automation(if not AI directly) will replace a good number of jobs and there isn't anything that can stop that. The argument against this is usually looking at past events like the industrial revolution where many factory workers were replaced with machines that were more efficient and accurate than humans.This opened up a ton of new jobs for humans to fill such as administrative or creative work. I think we'll have a similar shift and some new jobs will open up but no where near the amount of jobs that are displaced.