In the immortal words of Scooby Doo, "Ruh-roh."
I love robotics. I love AI. In theory. In the lab. I love good intentions, too, despite the fact that the Road to Hell® is paved with them. Ready for prime time they are not (to paraphrase Yoda).
The first "Whoops!" or "Doh!" moment involving an autonomous battlefield AI will make Sherman's march to the sea or Genghis' Holiday in Asia look like a junior varsity football game. If you're an optimist, you probably believe that we can recover from that. You might be right.
At least one of the commenters on the Guardian article made a point of the Cameron Principle. jebus1 said, "Have these people not seen Terminator?" Is James Cameron a prophet of doom of the same caliber as George Orwell? Many of Orwell's predictions were dead on the money, and it's starting to look like Cameron might be able to chalk one up, too.
Speaking as a recovering software developer, I personally believe that we're a long, long way from the point where we can rationally turn over ethical life-or-death decision-making power to machines... It's bad enough that we let humans make those kinds of decisions!
On a completely unrelated note, several people have mentioned to me how eerily similar our new storage cloud product name (SkyDrive) is to SkyNet. Try it out! Maybe the Robot Overlords® will value your allegiance and spare you in the future... (That last bit is funnier if you've seen the CTRL+ALT+DEL DVD.)