The Most Human Human

| August 7, 2011

The Turing test is one measurement of artificial intelligence: how closely can a computer program imitate human conversation? The test involves judges engaging in conversations with both humans and programs, both via computer, and if the judges cannot reliably distinguish the machine from the human, the machine has passed the Turing Test. In the book The Most Human Human (2010), author Brian Christian is a participant in the Turing Test as one of the human contestants, competing against computer programs, trying to convince judges that he is a person and not a machine.

Brian Christian meditates on communication and human interaction. What is a definitive characteristic of human conversation? Is it humor? Cultural references? Consistency in style and perspective? The author explains some of the most successful computer programs that have convinced judges in past competitions of humanness. One program, developed by the creators of the site Chatbot, has the remarkable ability to answer questions humorously; it also can give on-point cultural references (i.e. if one types in lyrics to Queen’s “Bohemian Rhapsody,” it will reply with the next lines of lyrics). Chatbot draws upon a large database of past human conversations; when you converse with the program, it is piecing together hundreds of thousands of prior chats, blended together in a conversational puree. So you are, in a sense, chatting with real people; at least the ghosts of real people, or the echoes of past conversations.

In-person, face-to-face, real-time communication goes so much beyond words, however. It is estimated that only 7% of human communication occurs on the verbal level; pitch, volume, and other vocal intonations comprise of 38%; and body movements (gestures and facial expressions) make up 55% (Koneya & Barbour, 1976). Communication is multimodal; it is visual, tactile, audible. So, what happens when inhabit an online social world of 7% communication? [Some writers such as Jared Lanier and Douglass Rushkoff have argued that this thin level of communication, as well as pseudonyms and anonymity, has dehumanized the Internet].

While I really don’t believe avatars or a Second Life-type of simulation will ever take over the average user experience of the Internet, it is interesting to think about how communication over computers will evolve over time. How will it go beyond the 7%? [And not just through the evident channels, such as live video chat]. How will it become more human?