HELIOLITH

One of the curses of modern day technology is the amount of time spent interacting with complex computer generated telephone routing systems. Clearly it must be cheaper to route and automatically place on hold thousands of callers each day than to listen to their ire and personally usher them forward to their destination.

I have a friend who is a master of the dying art of skillful reception. His style reminds me somewhat of the statue of the dancing Shiva, as he remains focused and calm while carefully, cheerfully, sensitively guiding calls to their appropriate connections.

The antithesis to this is a robot (bot) that can’t recognize what you’re saying, gives you the wrong choices, puts you on hold for a very long time, and disconnects you abrubtly, inevitably at the end. You have no recourse whatsoever.

The probability of… calling back and pinpointing, then gaining access to, the person that made the decision to fire the humans and hire the bots is quite slim.

Nonetheless, something I hadn’t counted on was the use of semi-intelligent chatterbots for customer service online. Chat windows are cheaper to run than phone based support, as you pay no additional connection fees, and agents can have many many windows open at the same time, so they are interacting almost simultaneously with several customers.

This entry posted yesterday by Mark Frauenfelder over at Boing Boing seems to my layperson analysis to be an obvious chatterbot. Just go read it fast, it’s short!

Okay, did you read it? It’s ridiculous right? No way that’s a human being is there? A technorati search on other blogs commenting on the same post reveals a few people who have come to the same conclusion (Dan of Dansays.com), and mikel at mikel.org who asked, “would this person pass the Turing Test?”, and a few others. It should be possible to create some sort of inverse Turing test that human’s couldn’t pass, but chatterbots could, but that used solely social dialogue as the measuring stick.

Chatterbots are quite cool, unless of course you’re trying to accomplish something in the real world. The oddity that I discovered at the Earthlink Website is that “EarthLink’s Live Chat is available 7 a.m. to 2:45 a.m. EST”, which either means they are using real people who work normal hours, that they are using chatterbots and trying to make them work normal hours so as to appear human-like, or maybe they use supervised chatterbots, and this one’s supervisor just decided to take a coffee break. I’m curious.

Boing Boing actually tipped me off to a few other interesting chatterbot factoids including, “When bots collide, Elizabot passes sex-chat Turing test, and points us to where, “Richard Wallace’s wonderful chatterbot ALICE” lives at http://aimovie.warnerbros.com/ go there, enter flash and you can test a chatterbot yourself. See if you can get Alice to sound smarter than the Earthlink chatbot.