When chatbots talk to each other, the conversation gets interesting in a hurry. Cornell University researchers rigged up a chatbot system to allow chatbots to talk to each other. The chatbot-vs.-chatbot interaction ranged from childish taunts to pseudo-metaphysical blatherings.
Humans who converse with chatbots often get frustrated with the chatbots' seeming stupidity and inattention. Watching a couple of chatbots get snippy with each other for being stupid and inattentive is quite entertaining and satisfying. (See the video below.)
The chatbot-vs.-chatbot avatars are a British man and a South Asian woman, both instances of Cleverbot, developed by artificial-intelligence programmer Rollo Carpenter. The software has learned phrases from millions of conversations it has had with humans on the Internet.
At one point the male Cleverbot declares itself to be a unicorn. At another, he tells her she is unhelpful and therefore a "meanie." She dazzles him with her philosophical prowess, declaring that not everything could be half of something. My favorite part, though, was when one bot threw bot-ness in the other's face. The male says, "You were mistaken. Which is odd, since memory shouldn't be a problem for you."
Cleverbot won the 2010 British Computer Society Machine Intelligence Competition. The ultimate in chatbot competitions is the Loebner Prize Competition in Artificial Intelligence, a $100,000 contest to see if a piece of software can talk its way into convincing human judges that it's human. A chatbot winning this prize will be a momentous event.
As IEEE Spectrum's Evan Ackerman notes, the first computer program that can carry on a conversation indistinguishably from a human is arguably the first computer program to demonstrate an artificial intelligence.
Even if we eventually manage to make computers truly conversant, I hope somebody keeps a few these not-so-clever bots around. They are entertaining.