A telemarketer who is apparently pretty persistent in contacting people to sell them health insurance… turns out to be a robot. From the sound of it (there are two audio recordings), what we have here is a computer with decent speech recognition software and a “dictionary” of pre-recorded responses. I believe an actual human read from what was probably a pretty extensive script, and the computer selects from a range of appropriate responses. However, there are some pretty simple ways to trip it up. Ask it if it’s a robot, and it will tell you that it’s a real person. But ask it to say “I am not a robot,” and it won’t… probably because it can’t. The list of recorded clips doesn’t contain that, so it can’t say it. Similarly, questions like “what vegetable is in tomato soup” and “what day was yesterday” throw it for a loop. The fairly consistent delay between confusing question and roboanswer is also a good indicator

I do wonder about the legality of a telemarketer, human or computer, that will tell you an outright lie.

Best not to be too complacent about human supremacy here. Sure, a few well-chosen question and you can confound this robotelemarketer and determine that it is, in fact, a robot. But most people *won’t* go to the bother; at first listen, it’s pretty convincing. And  give it just a few years, and a second-hand smartphone with Suri Version X will pass even a rigorous Turing test. So if your job is to deal with people over the phone, your days are numbered. How long before other sales jobs are turned over to bots? Sure, it’s a hell of a step from “voice on the phone” to “robot in the showroom,” but who knows. I’m pretty sure that burger flippers could be relatively easily replaced by machines. And now the person running the cash register can be easily replaced.

It’ll be curious to see how all this settles out a few generations down the line when *every* human can be cost-effectively replaced.