There’s a nifty piece that went up on Salon a few weeks ago about Hugh Loebner’s annual Turing Test competition. In case you’re not familiar with the competition, it’s an annual contest where computer scientists (and hobbyists) vie to see who has the most life-like chatbot — whose program can converse in the most human-like fashion.

The writer noticed that “serious” A.I. researchers — from universities — typically deride the Loebner contest. Their argument is, usually, that chatbots aren’t real A.I. When you type a sentence in to a chatbot, it doesn’t “know” what you’re saying; it just looks for a preprogrammed response that matches the input, and spits it out. Chatbots usually don’t learn. So serious A.I. scientists regard them as nothing more than a trick. In fact, they often don’t even deign to enter the competition — because they say that the test isn’t tough enough:

“Well, I’ve given you the reason,” replied Dennett. “Think about it: If you and your lab/team had devoted years to developing a truly competent language-comprehension system, but it could be beaten by somebody’s cheezo hobby system because the rules didn’t permit putting a real strain on the competitors, you wouldn’t enter that competition. You wouldn’t take that competition seriously. You don’t enter your Ferrari in a ‘race’ to the bottom of the mountain that can be won by the first car that drives over the cliff and lands upside down on the finish line.”

Behind all this skepticism is one powerful, common-sense assumption about humans: That our conversation is sufficiently complex that it can’t be captured by a “dumb” chatbot. Fair enough.

Except for one thing — many humans can’t converse in an intelligent manner. A.I. chatbot-maker Richard Wallace made this point when I profiled him last year. Sure, the conversation offered by a chatbot is rambling, disconnected, illogical, and unable to follow a topic for more than a sentence. But again — that pretty much describes most conversation online. “Go to any AOL chat room,” Wallace once urged me. Read the stuff that actual humans are writing to each other. It’s easily less human that the speech coughed up by even the dimmest chatbots.

Indeed, here at Collision Detection we have examples even closer at hand. Back in December, I wrote a little posting about the infamous “Michael Jackson Baby Drop Game”. Soon, Google picked it up, and a search for “michael jackson baby drop” began listing my post as one of the top results. Thus, Michael Jackson fans started dropping by — and writing their comments.

And the thing is — more than half the postings are wildly, gorgeously incoherent. You can check ‘em out here, but here are a few samples:

- Stop the negativity because maybe he was bapting them. I think his wires are crossed to but he acks like he is under some kind of religion that he sat for himself and his kids. They have to cover their faces in public.

- YOUR SO FUCKING OUT OF YOUR MIND MICHAEL NEVA EVA MENT IT LIKE THAT THE PEOPLE BELOW WERE BEGGING TO SEE MICHAEL ,SO HE DECIDED TO SHOW THEM ALL HIS BEAUTIFUL CHILD AND THE FUCKING PRESS HAS TO DO THIS???

- Michael, I luv ur music but ur a FUCKIN idiot!U almost dropped that baby.

- you are awsome but what the hell mike you have screwd it up for yourself,it cant end good now iluv ya but get it strait dude, later king you are the best

- hi im very fan of michael jackson but some times his stupid

Interestingly, the posts seem to get more incoherent as time goes on, as if there were some subtle half-life-decay algorithm of moronicity at work here. Even more alarmingly, the more blitheringly weird the posts become, the more likely the authors are to address the post to Michael Jackson himself. Yeee.

I could be accused of poking fun at people for being illiterate, or merely crappy at typed/verbal speech. That would be a correct assumption. But I do this in the service of the larger point that A.I. researchers frequently miss: We humans are defined as much by our idiocy as by our brilliance.