Robots and human conversation

Brit programmer wins chat-bot prize

From the article:

I guess we have a long way to go before chatter bots seem human. The conversation transcripts for George (Jabberwacky) are obviously not human. However, George seems to be as good as IT help lines?
Computers are stupid because they only answer questions. :)
My question is is Jabberwocky improving primarily from his AI, or by programmers analysing his flaws and rewriting code?
From all that I have read in the last few years Jabberwacky is a canned response program. It is preprogrammed to anticapited questions made by users. I prefer the chatterbot program call Ultra Hal Assistant. It really learns from interaction with its users. It has the ability to make assumptions from different data that is fed into it by conversations with users. If you have not tried it out they offer a free version with almost all the same power as thier commercial version. It can be found at Zabaware.com
If you like Jabberwacky, you will love Questsin. More tricks, more kicks! Just add questsin@hotmail.com to your MSN Messenger buddy list and chat away.
Post a Comment

<< Home
Archives © Copyright 2005 by Marshall Brain
Atom RSS

This page is powered by Blogger. Isn't yours?