Page 106 - Socially Intelligent Agents Creating Relationships with Computers and Robots

Social Intelligence for Computers 89

violence of exchanges and the extreme positions participants take, one could
say that the goal of exchanging ideas has been subsumed and remains merely
to sustain the appearance of a meaningful social sub-system. Most of the time
the type of interactions involved and the shape of the communication mirror
classic power relations found in the "real" world.
When teenagers interact by participating to games over the Internet, they
know very quickly who is reliable or not and sometimes, they choose to play
a session just because they know one participant and want to play with him
(rarely her). While playing, they can identify the type of behaviour the other
person has and form a definite opinion of it, e.g. like it or dislike it. Even if
they don’t actually meet the people, they often mix relations and communicate
more intimate information through ICQ, give advice to each other about how to
find music, new free games, or videos. The relations are usually more flexible
and less passionate than those of adults, maybe because younger people feel
more comfortable with the computer tool itself, or because the relations they
experience in real life are more relaxed as well.
The pattern of relations and the rules of communication develop in ways that
are similar to that in the outside world, but not identical. They also constitute an
independent history of computer network communication that is very creative
in terms of language and social codes. Existing structures are important as they
provide the foundations for interaction and a form of justification for it, but
the actual establishment of alliances between individuals is mainly based on a
new ability: the ability to use the norms of that special network to distinguish
oneself or identify others.
Two elements that are at the basis of human socialisation can be recognised.
An important ability of humans is the recognition of the other as a "self", similar
to his own - this makes it possible to anticipate or interpret the other’s actions
by projecting one’s own acts [19]. Secondly, humans can create new norms
of communication in sub-groups, mainly by adapting norms that they already
know: the mental environment that enabled them to learn how to communicate
and be creative [21].
Identifying a person as being relevant for communication has not been suc-
cessfully implemented in AI. It is closely related to the idea of creating an
artificial consciousness - one that would be able to identify a meaningful event
without having preconceptions on what to look for but only an ability to learn
[6]. Since that kind of artificial system is not available yet, most of the work
about "like-me" test [7] postulates the ability to recognise a relevant self. They
develop the idea of misconceptions in projection that can lead to the creation
of a strange social life, quite unpredictable, but capturing essential properties
of human societies [8]. Others try to teach useless robots how to manipulate
humans in order to keep themselves working: one can hope that by postulating
this surprising egocentrism, the robots will build strong representations of their