Molding Cleverbot

[YourEverydayWaffle]

Hey all. I've been wondering, is there a way we could make the Cleverbot AI "predictable?" For those of you who don't know, Cleverbot is an "AI" (more of like how a parrot functions, repeating phrases said by humans earlier based on what the human said now, not really much of an AI sort of thing) that is supposed to mimic human-like conversation. You start by typing something in, usually a phrase like "How are you," and the bot "responds" based on what a human said when it said the phrase you did. So, bot says "I smell like raisins," which, of course, the logical response to which would be "As do I." The bot will remember that response and might use it in the future the next time a human says "I smell like raisins" to it. So, anyways, any ideas on how one could make the bot's response constant, the same, every time you say a key phrase? I was thinking that I could write a simple .vbs script to type in a sentence, hit the respond button, write another "response", refresh the page, and repeat, however, this wouldn't work, because it would take a ridiculous amount of time and it would tailor for multiple responses. Any ideas on how one could do this, preferably with some sort of repetitive script in this case?

[Souper]

Type in "What happened to Ben?"

The answer will usually be the same every time.

[YourEverydayWaffle]

Type in "What happened to Ben?"

The answer will usually be the same every time.

You're right! Even better for those of us who get the reference.

[TechnicalK3rbal]

I asked it, "Are you human?", and of course it answered yes, since that's what the normal person's answer was. I then spent 10 minutes trying to convince it that it was a computer program, and then felt stupid.

[MC.STEEL]

I am more curious about giving it the ability of creating its own personality and not remember what the general public responds.

There could be a file with parameters like:

-does not like to swear.

-likes science.

-can be a jerk at times.

The problem with this is how do we program it so those parameters mean anything.My idea is linking it with an online dictionary (just like Watson) and search the description for keywords.If it finds that the word the user mentioned is linked with science for instance (say neurology).It will say "brb googling neurology in " because it contains the keyword science and thus interests it.(it will also remember any data gathered from the search)

Alternatively the machine could just ask the user what it is.If the machine becomes confused with the explanation it will start questioning is it that or is it this,to witch the user responds with a simple yes or no.

Over time the machine will learn stuff that interests it and what doesent.Giving it the impression of character depth.

I can see this with improving digital companions (like Siri) and even be used with video game technology,giving the player the ability to speak to Characters by typing text and not just to use the severely limited dialog tree.

Overall i see this as the next logical step.We could possibly learn a lot from how children develop tastes and preferences on their own.

Edited April 25, 2014 by MC.STEEL