Post 9: Some wild thoughts about AI vs. human communication late at night

A high school classmate, now a professor of computer science at a prestigious university in Shanghai, always touts in his blog how amazing artificial intelligence is and predicts that computers are going to beat human beings in intelligence and replace humans in the near future given the progress that has been made in computer science. Well, that’s because he just takes human intelligence too much for granted. If he had ever realized how sophisticated human communications and interactions are, he wouldn’t have said that. If he had only read Sidnell and Stivers!

Sidnell (2010) and Stivers (2008) lead me once again to delve deep into the wonder world of social reality humans create for themselves by story telling and understanding. We are telling each other each day stories that are understood, interpreted and responded but seldom are we conscious of the intricate pattern and design in our narratives that expose our own stances and involve and elicit from those of our listeners. Sidnell reveals to us the way in which a story is usually occasioned, either by a question, a local stimulus, or a prior story in the context. With a story-preface, the teller provides the recipient with clues as to what it takes for the story to reach completion. Unlike a robot, the listener can either offer a preferred or dispreferred response upon the completion of the story based on his recognition of the teller’s stance revealed and reported in the telling. Furthermore, Stivers unfolds to us the intricate mechanism through which a teller’s stance is unmistakably conveyed to the listener, whose shared ability to decipher these devices is something the teller depends on for his design of the story. To me, it is really amazing to know that even the average grammatical devices such as voices, aspects, tenses, determiners, and the gerund, etc. are not as innocent as they appear to be. They are all possible resources skillfully employed by the teller, offering the listener access not only to the event narrated, but also to the stance of the teller. Equally amazing is the listener’s ability to understand the stance demonstrated through an intricate system of recognition: either by alignment or by affiliation, the nuance of which is the highlight of Stivers (2008). By vocal continuers, like um hum, or yelp, the listener aligns himself with the teller by semantically acknowledging the reception of the information in the telling and by structurally supporting the asymmetry of the format of the story telling. However, by nodding, which only shows itself when the teller reveals his stance, the listener affiliates himself with the teller by endorsing the teller’s perspective after his achievement of the access to the teller’s stance.

A side effect from reading Stivers is that it proves my long-time hunch that doing CA is like doing psycho-analysis, as what we used to do in psycholinguistics by looking into linguistic devices for evidences. I happened to find that Stivers is affiliated with an institute for psycholinguistics. This reminds of me of Jessica’s suggestion that I should take good advantage of my linguistic background. Now I come to see where to seek the bridge. Still, I can’t see a future in which computers will be as smart as humans and can talk to me the way a friend can.