Book Reading #48 - Media Equation

Title:
Part 1: Machines and Mindlessness: Social Responses to Computers
Part 2: Computers are Social Actors
Part 3: Can Computer Personalities Be Human Personalities?

Reference:
Nass, Clifford and Moon, Youngme.  Machines and Mindlessness: Social Responses to Computers.  Journal of Social Issues, Vol. 56, No. 1, 2000, pp. 81–103.  http://onlinelibrary.wiley.com/doi/10.1111/0022-4537.00153/pdf

Nass, C., Steuer, J., and Tauber, E.  Computers are Social Actors.  CHI '94.  http://delivery.acm.org/10.1145/200000/191703/p72-nass.pdf?key1=191703&key2=6181032031&coll=DL&dl=ACM&ip=165.91.4.172&CFID=17171862&CFTOKEN=58852860

Nass, C., Moon, Y., Fogg, B., Reeves, B., and Dryer, C.  Can Computer Personalities Be Human Personalities?  CHI '95.  http://delivery.acm.org/10.1145/230000/223538/p228-nass.pdf?key1=223538&key2=9381032031&coll=DL&dl=ACM&ip=165.91.4.172&CFID=17171862&CFTOKEN=58852860


Summary:
Part 1: Machines and Mindlessness: Social Responses to Computers

Nass goes over people and their social responses to computer.  He conducts four studies which will be discussed in this section.
The beginning of this experiment was similar to experiment 4 is "Computers are Social Actors" (Part 2).

The experiment consisted of three rounds of questionairres and responses from computers with either a male or female voice.  The users filled out a written questionairre at the end cataloging their perception based upon the gender of the voice.  It was found that male voices elicit a greater sense of correctness and likability.  Males are also found to be thought of as more intuitive on technological matters whereas females are on relationships.

After gender, the next variable studied was ethnicity in people's responses to computers.  This tested "mindless stereotyping" because upon being asked a situational question based upon risk the researchers saw if there was a difference when the users spoke to a face of a Caucasian or Korean on the video screen.  It turns out this was the case. 

The third section of this experiment tested whether people would have different responses to different groups - one they were in and one they were not in.  Participants were more likely to cooperate with the computer when it seemed to be representing the team of the user. 

In a different experiment, it was found that people are polite to a computer.  It was found that the reciprocity norm was also held, that the user helped the computer that helped it and did not help as much the computer that did not help that person as much.  An extension of this looked at how people divulge intimate information.  The more likable reciprocity the computer the more likely the person was to divulge more intimate information to it. 

Another series of experiments dealt with perception.  There was a generalist and specialist program and they found that people thought the specialist program was more informative. 

The next section addressed possible explanations for the above experiment results.  The first addressed was anthropomorphism.  This was immediately rejected because the people being studied were adults and denied that they thought the computer was a person.  The next objection is that the people actually are intending to respond to the person behind the computer, or the programmer.  First, when questioned the users did not think of the programmer behind the computer.  Next the users thought that all computers were programmed by the same person, eliminating the bias of different people behind different computers.  Finally when the terms "computer" verses "programmer" were used, computer was more favorable to programmer.  The final argument against these experiments is that the questionnaires made the users forget that they were dealing with a computer.  Nothing was fancy about these that would motivate a user to elicit human attributes or think it wasn't a computer.


The last part of the paper was future research described here.  They wonder what the characteristics of the computer are that make users make these attributions to computers.  They go through hypotheses of these questions.  They also discuss similar experiments that they did as compared to a real human taking the place of the computer. 

Part 2: Computers are Social Actors

This paper seeks to show that people are not expecting their interaction with a computer to be the same as that of a human.  In order to do this, this study looks at five questions and studies each one.  The questions include

1. "Will users apply politeness norms to computers?"
2. "Will users apply the notions of ‘self and ‘other’ to computers?"
3. "On what basis do users distinguish computers as ‘self or ‘other’ — the voice or the box?"
4. "Will users apply gender stereotypes to computers?"
5. "If people do respond socially to computers, is it because they feel that they are interacting with the computer or with some other agent, such as the programmer?" and, "Who or what do users think of when the a computer says ‘I’?"

In the experiments set up, each subject was to be instructed by a computer tutor.  The outcomes of the experiments are listed below.

Experiment 1:
Here a paper and pencil questionnaire was compared to a questionnaire asked by the computer.  It was found that subjects claimed the computer was more helpful in a tutor than pencil and paper.  
Experiment 2:
Here there were two computers used in order to distinguish between two entities where one was critical and one gave more praise.  It was found that the user did treat the computers differently.
Experiment 3:
This is an extension of 2, but here there was a fluctuation of voices spoken to the subject of the praise and criticism in order to get a more general outcome.  It was found that the subjects respond to different voices as different social actors.  
Experiment 4:
In order to do this experiment, the researchers altered the voices coming from the computers in the training time to either male of female.  This experiment tested perceptions of male praise being more likable than female praise which was found true, females being more knowledgeable on relationships which was found to be true, and praise from males is more convincing than females which was actually found to be relative to the style of the subject.
Experiment 5:
Here the subjects went through the rounds of the experiment twice with different labelings of the computer and the experimenter (I, the computer, the programmer).  It was found that the use of "I" was liked over "the computer" over "the programmer".

The paper ended by summing up the findings and extrapolating that among other things CHI is social interaction.

Part 3: Can Computer Personalities Be Human Personalities?

This tested submissive and dominant personalities of computers upon users interpretations.  For the experiment, subjects were either paired with a submissive or dominant computer.  The subjects then answered a questionnaire.  The results found that the people could tell which version was more dominant, and people liked the similar personality to their own in the computer showing that they did judge the computer based upon its personality. 

Discussion:
Part 1: Machines and Mindlessness: Social Responses to Computers

This paper had no pictures.  So in following Paul's lead, I was very disgusted while reading this paper.  But aside from being difficult and fairly boring to read, it was nice to get the overall theme that people interact with computers in similar ways as regular people.  This could have been summarized in at most a three-page paper that would have been so much more interesting.  

Part 2: Computers are Social Actors
It is laughable to take anything from the results from any of the five experiments.  They all seem like a joke in the context of anything larger than trying to participate in the scientific method, or a mock-research project.  I suppose the concept of people treating computers as people is fairly useful, but the other aspects and ways of conducting the experiment to me seem absolutely ridiculous.

The best part about this paper is the summary at the end where it has the bullet points under each of the experiments.  This is really all you need to read of the paper.  Unfortunately I read the whole thing (sad face).

Part 3: Can Computer Personalities Be Human Personalities?

PRAISE GOD FOR A SHORT ARTICLE!  This article was by far the best because it was the last and shortest.  However in terms of content it is not a novel idea that people could tell from a set of questions whether something, whether a human or not, could tell whether it had a dominant or submissive trait.  We can see that in animals and even art.  Way to go, Nass, you have proven yet another instance of extremely uninteresting facts.  Thank you, though, for making it only two pages.