I'm quite fond of this XKCD from last month:
I think AI has a lot of benefits to offer: they've proven IBM's Watson to be very helpful in medical diagnosis and preventing drug interactions, and that's definitely a good thing! I know pretty much zero about AI. I think it's an interesting field that I'd like to study some day, but I don't think I have the appropriate education background to go in to it seriously.
(and since I mentioned Watson, I have to mention a pet peeve: yes, it beat Garry Kasparov. I think a much better demonstration would be to see it in a round-robin tournament between 15 of the highest grand masters and see how it would fare. It was optimised to beat Kasparov, I'd like to see if it could generalize and adapt to defeat all of the others.)
I'd like to see AI applied to network security and packet inspection, where it actually is to some degree. If it can be trained to a certain baseline to recognize what a network attack is and to prevent it, or on the other end to notice what improper data exfiltration is and prevent that, it would be a tremendous boon. Such an appliance located on Target's network edge could have stopped that hack cold, but the tech isn't there yet.
[font="Arial"]Knowledge is of two kinds. We know a subject ourselves or we know where we can find information upon it. --Samuel Johnson[/font]