Man versus machine
Attempts to develop speech-recognition technology appear to be based on the questionable assumption that nearly every caller would opt to speak to a real person rather than a machine (Hello, HAL, “by John Searbrook, June 23rd). However, an infinitely patient machine can be available twenty-four hours a day, every day, and it poses no social risks, such as self-consciousness or embarrassment. In a test we conducted, pitting machine against real-person conversation, participants completed a brief personality survey and were then asked to choose one of three ways to organize their thoughts about their survey experience: interacting online with participants, interacting with chatterbots, or private typing. We found that while most of our participants – forty-two per cent – chose real person interaction, thirty-eight per cent chose machine (chatterbot) conversation. Follow-up experiments and analyses showed that the machine preference was stable and could not be attributed to participants’ mood, to their perception of textmessaging difficulty, or to the novelty of a chatterbot. Some of us would rather stick to automation, whatever its limitations.
(The New Yorker, August 4, 2008, p. 5)
Follow-up experiments and analyses showed that
