Let’s face it, the Turing test is easy. OK, so no computers have actually passed the gold standard yet, but it’s only a matter of time before somebody programs one with enough recorded chat up lines and put downs to convince someone it’s real. I’ve got an idea for an effective new test of computer AI: it’s called the Sillis test. Yeah, I named it after myself.
How does it work? Simple. A robot has to be smart enough to create its own identity, profile and following on the world’s most popular social networks. And I’m not just talking followers and friends – they’re easy to acquire, even for computer programs – but real authority, as measured by organic factors like retweets on Twitter, or the number of people in Facebook groups it sets up.
Read More