Page 102 - Socially Intelligent Agents Creating Relationships with Computers and Robots

Chapter 10

SOCIAL INTELLIGENCE FOR COMPUTERS

Making Artificial Entities Creative in their Interactions

Juliette Rouchier
GREQAM (CNRS)

Abstract I review two main principles that have been developed to coordinate artificial
agents in Multi-Agent systems. The first is based on the elaboration of complex
communications among highly cognitive agents. The other is eco-resolution,
where very simple agents have no consciousness of the existence of others. Both
approaches fail to produce a social life that is close to that of humans, in terms of
creativity or exchange of abstractions. Humans can build new ways of commu-
nicating, even with unknown entities, because they suppose that the other is able
to give a meaning to messages, and are able to transfer a protocol from one social
field to another. Since we want social intelligence to be creative, it seems that
a first step would be to have agents be willing to communicate and know more
than one reason and way to do so.

1. Introduction

Here, I compare computers’ social intelligence to the human one. There is
no generally agreed definition of social intelligence, but several elements seem
to be indicated by the difference between human intelligence and more basic
cognitions. These include: the ability to communicate with others in order
to undertake common actions; the ability displayed by a society to integrate
newcomers (and conversely for individuals to adapt to new ways of interacting)
in order to communicate with unknown people; the ability to understand what
others want from you, how you can help, or conversely influence others so that
they help you [1].
Some progress has recently been made towards the understanding of social
intelligence for non-living entities in the field of Multi-Agent Systems (MAS).
MAS focuses on the socialisation of artificial intelligences using a variety of
approaches. Attempts to create a real artificial intelligence are often based upon