Do Android Trolls Dream of Electric Bridges?

Guys, I think everyone’s favorite troll is a bot. Jeff, do you have that bit of robot, webcrawler, and bot blocking text in your header? I put it in, and he doesn’t seem to come around. (That and I’ve got him banned from commenting in haloscan.) How many of you have it in your header? I was getting hits from him, but haven’t, once I fixed a problem with the header. (I mis-typed something.)

He acts a lot like one of those AIM sex bots. I mean, think about it. once you get to a certain point in the conversation, if you keep responding, he’ll respond with something completely off topic, but a previous topic of conversation. It doesn’t matter what you respond with, so long as certain phrases are there, he’ll respond with a set (pre-programmed?) response. But if you get to something that breaks the query/response loop, you’ll get gibberish, like it’s some sort of parse error.

For instance, Eliza. Here’s another interesting article from Salon. (Would seem to contradict my theory, since humans don’t always pass the Turing test.)

University of Tulsa does have some people doing AI at it. I could see some grad student getting ahold of the book our favorite troll supposedly wrote for a couple of bucks and using that to set up certain facts about the bot’s “life.” Eliza’s decades old, and I know advances have been made in the field.

Then again it would be more likely (and more ethical) to have a bot you wouldn’t notice. It responds appropriately to threads, and goes on its merry way without leaving a shitstorm in its wake. Unless it is learning from analysis of threads. The internet’s a nasty place, so I could see any AI loose on it going psycho.


~ by Jen on February 21, 2009.

%d bloggers like this: