Karan Sharma
The paper Designing Knowledge Based Systems as Complex Adaptive Systems said
This paper proposes that knowledge based systems must be designed as complex adaptive systems and any other approach is not fundamental, even if it sometimes yields good results. Complex systems are characterized as having global behavior not always explainable from local behavior. Here we propose that the way we perceive knowledge in AI needs to change to Complex Adaptive, hence the need for a paradigm shift is stressed.
Almost all historical KBS were not complex systems in an authentic sense. But it is not a good idea to criticize them because with available resources and theories, they did their best. Sooner or later, we will have to design our KBS as complex adaptive systems, so why not sooner. There are three mechanisms that must be part of any knowledge based system, viz., Interdependency and fluidity, mechanisms for attribution of emergent properties, and self-organization.
Karan Sharma was the author of this paper which will be
presented
at
The First Conference on Artificial General Intelligence
and
is pursuing his M.S. in Artificial Intelligence at the University of
Georgia. Currently, he is working on theoretical foundations
of Fluid Reasoning and Representations that will aid in the creation of
Artificial General Intelligence. He strongly believes that the time has
come to renounce traditional cul-de-sac rigid AI in favor of more fluid
and plastic approaches.
His research interests include:
- Complex Systems: Emergence, Self-Organization, Adaptive Agents.
- Evolutionary Computation: Genetic algorithms, Genetic Programming, Theory.
- Natural and Computational Creativity: Analogical Reasoning, Complex Systems Approach.
- Artificial General Intelligence: Knowledge Representation, Information Theoretic Approaches, Theory of General Intelligence.
- Others: Evolutionary Psychology, Cognitive Fluidity.
