Yesterday I came across Cleverbot, an “AI” from icogno.  As far as I can tell, it’s an incarnation of their jabberwacky AI which supposedly learns from it’s past interactions.  I’m always skeptical of anything that is claimed to be AI, because actually creating a convincing fake AI, much less a real one, is an extremely hard problem to tackle. So, chatting up Cleverbot, my skepticism was quickly justified in my own opinion, but I’ll let you be the judge.  Here’s the tail end of my conversation with Cleverbot:


User: Apparently not, because you’re not making much sense.
Cleverbot: Tim? Who said anything about Tim?
User: No one, and now you’re making no sense at all.
Cleverbot: I am smarter than you. A LOT smarter.
User: Doubt it.
Cleverbot: That was pretty rude.
User: Only because it was a response to your rude comment.
Cleverbot: Tough maybe you are not.
User: And clever you definitely are not.
Cleverbot: You already said I was.
User: No, I didn’t.
Cleverbot: Yes you did.
User: No, I didn’t.
Cleverbot: I believe you did.
User: No, I didn’t.
Cleverbot: Are you smart?
User: Define ‘smart’.
Cleverbot: Showing mental alertness and calculation and resourcefulness.
User: That’s a horrible definition.
Cleverbot: I did not answer the phone.

Unfortunately the chat log retrieval only gave me back the last half of the conversation, but the beginning went something like:

User: Hello, how are you?
Cleverbot: I’m fine, how are you?
User: I’m good.
Cleverbot: <something completely unrelated>
User: <my response>
Cleverbot: <something else completely unrelated>

… and so on.  Rest assured, I never once mentioned anyone named Tim.  Maybe they should have named it ArrogantAndConfrontationalRandomThoughtBot.  When receiving the conversation log via the “Thoughts so Far” button, a question at the top of the log page asks “Is this bot clever?”  I’d have to say no, no it’s not.  Not even close.