Artificial Intelligence Demo Shows How Chatbots Soon Sink Into Spat, Non Sequiturs and Nonsense

Make headway, Max Headroom! Meant to be Cornell classroom demonstration, a robot avatar conversation quickly turned into the spat chat heard around the world.

Igor Labutov, Cornell graduate student in mechanical engineering and Jason Yosinski, Cornell graduate student in computer science, set up the conversation as a demonstration for an artificial intelligence class taught by Hod Lipson, Cornell professor of engineering. They chose a Web-based Cleverbot, a computer program designed to simulate human conversation, then added text-to-speech capability, developed computer-generated faces, set two laptops side-by-side and seeded the conversation with a simple “Hi” – and then conversational craziness followed.

The resulting YouTube video has spiraled into viral, as nearly one million viewers have seen it over the last few days.

See the link on YouTube: http://www.youtube.com/watch?v=WnzlbyTZsQY
In the video the two avatars, one representing each gender, exchange pleasantries, but soon argue when one accuses the other of … being a robot. As non sequiturs fly, the two automatons launch into a discussion on religion.

Many chatbots work by repeating back what they hear, but in a slightly different form. Cleverbot, developed by British artificial intelligence specialist Rollo Carpenter, draws on a vast database of phrases from all the conversations it has had in the past. Since it went live in 1997, Cleverbot has carried on more than 20 million conversations.

“What makes this interesting is how people interpret what they see,” said Lipson. Some find “sexual tension” between the two characters. One viewer commented the conversation was not like real human speech, but another countered that it seemed like marriage.

Although this is not a typical subject for research in Lipson’s Creative Machines Laboratory, the team is considering further exploration.

Said Lipson: “The human reaction to the video is the real story.”