Am I the only one who doesn't 'get' AI? I have a book on the subject, and although it is interesting, it struck me that all AI appears to do, is analyse data etc. and act on it, faster than man can. The book talks about finding patterns in data that we might not spot, and thus unexpected solutions to problems, but again it strikes me this is a matter of speed, rather than 'intelligence.' Given enough people and time it seems to me that anything AI can do we too could, albeit much slower. For me, this doesn't make AI intelligent, just fast.
The other aspect I don't understand is the issue of AI 'learning' based on 'reading' existing documentation, so that it can absorb huge amounts of information and can use this in any tasks it is undertaking. How does it sort the wheat from the chaff? How does it work out how the rantings of Donald Trump compare to the utterings of any other US president, or how it can differentiate the absolute rubbish put out by some people online, compared to those who actually have a sensible argument? In short, how does it sort the truth from the lies? Plus, of course, how does it learn morality, when in literature there are fine lines which are often blurred. In short, how can we ensure that AI doesn't decide that the problems of the world are caused by man, and decides to do away with us? After all, this would likely solve the problem of climate change, which it will have learned is a major issue at the moment!!