|
EPIA'03 - 11th Portuguese Conference on Artificial Intelligence
MAAII -- Multi-Agents and AI for the Internet
|
Session: December 6, 14:45-16:15, Room A |
Title: |
Towards Individual Power Design --rediscovering the will of acting agents-- |
|
Francisco Coelho and Helder Coelho |
Abstract: |
We started the setup of a virtual laboratory for the
study of power struggles where agents embody different
capabilities, skills, tactics and strategies. The
scenario is a caricature of hierarchical production
societies and are driven to be structured into
different organisational forms. Under the assumption of
limited rationality, we look at mechanisms able to
drive the choice of action and to act. On our simple
setup, an agent will look to change to a new leader
(flee) in a poor present individual situation. On the
other hand, if the agent is confortable, it might
(become attractive and) lower the income tax imposed to
it's subordinates.
Our agent models are yet too simple to distinguish
motivation from will: the decision procedures aren't
sophisticated enough to enable the consideration of
reasoning processes running at each action choice. Thus
we can't identify reasons (or: motives) driving action
selection. However, there is a causal relation from
agent's state and it's choosing process (but not it's
action).
Can we say that each agent wants to survive? Their
constitution drives them to do so. Perhaps their
tactics aren't good enough and their actions
unsuccessful. Perhaps they don't have enough power to
reach their goals. But, can't we speak here of will?
Special care is given to the validation of conjectures,
where data collected from simulations is subject of
statistical analysis towards their infirmation/refutation.
Our initial experiments point out that, in these
societies, 1) Life Expectation and Individual
Productive Capacity are related with the Hierarchical
Rank and 2) If, under certain critical circumstances,
agents can change to new leaders --or eventually become
one-- Global Productive Capacity reach higher numbers
but Life Expectation is lower. |
Back to schedule. |