There was a news report on our local NPR station about a company in the area that's improving computers' ability to recognize the content of messages.
Simpler systems see the word "love" and assume the message is positive, or the word "hate" and assume it's negative, and route it to what it thinks is the appropriate department for a response.
The new system can look at the message more holistically. I was cooking, so I might have missed something, but I specifically recall them saying that it could detect sarcasm.
Oh, crap. I figured that was our one insurmountable trump card when it comes to the apocalyptic battle between the machines and us.
"I love, love, love our robot overlords." [Fixes a charge to one end of the central brain machine.]
"I know, man, they're like, totally awesome." [Fixes a charge to the other end.]
...So I guess that won't work.
Time for Plan B: