I'm thinking of looking into using neural networks to design a chat bot, however I have herd that a system similar to neural nets was used for the previous loebner winner. Does any one know how they used the ANN?
This is from the Icogno website, which is the place to go for info about Jabberwocky (the 2006 Loebner winner): Our AI is not an Artificial Life model, nor a Markov Chain, nor a Neural Net, nor a Fuzzy Logic system, or any other single recognised AI technique. It is a complex, layered set of heuristics that produce results through analyses of conversational context, and positive feedback. Our approach is data-centric, probabilistic and statistical, yet at the same time distinctly chaotic in nature - tiny differences in context can make huge differences, as is true of human life.
Really, jabberwacky is based on something similar to ANN ? I never knew that, I hadn’t really taken it serious. All though I have found it by far the most funniest bot. They do actually say is not neural net, more algorithm based, but I don’t know of any chat bots based on neural nets, sorry Jay.
My AI Bot Quark is a static neural net running within the Zabaware AI Engine. I would post a link but unfortunately the system will not allow it, so you can get there by clicking my profile and going to my home Site. Funny to see a post in this thread from my best friend Richard Gray, aka KnyteTrypper. I still can't believe how far and wide he posted ! YOURS -- Christopher Doyon
i can see that the chat bot you crteated is very high tech...aisde from chatting, will it ever be able to analyze short stories?
