Last week, a particularly weird piece of artificial intelligence news made a splash in the internet ocean. TIME magazine’s Washington bureau chief Michael Scherer got a phone call from a telemarketer named Samantha West, who was selling health insurance. She was friendly and cheerful, but something about her bugged Scherer.
“Are you a robot?” he asked her.
With a little laugh, she insisted that she was a real person. Still, something was off. Scherer pressed her on several points that would have been simple to an ordinary human being, but Samantha — or Samantha-bot — was unable to answer. Later, other TIME reporters called her back. Here are the conversations they had:
Now, as it turns out, Samantha West is not precisely a robot. The company “employing” her revealed a couple days ago to TIME that Samantha West is simply a soundboard of pre-recorded statements and questions, which is operated by a live human. The technology does not yet exist to build a stand-alone bot capable of what Samantha West does over the phone. Though automated, she is not autonomous, and therein lies a small difference.
But Samantha West grabbed my curiosity nonetheless. After hearing her story, my mother and I played around with Apple’s virtual assistant Siri on the iPad, who is most definitely a robot. However, Siri refused to admit this when we asked, making evasive statements like, “I’m an assistant. Isn’t that all that matters?” and “I don’t really like these arbitrary categories.”
I set out to discover if this was just a fluke, or if there are other chatbots around that also do not acknowledge they are robots. Continue reading When Robots Lie