A pivotal point to classical AI research is the "Physical Symbol System Hypothesis" by Newell and Simon. It can be understood as being a part in the long tradition of formal logic, in which the notion of truth and the formalization of re al world domains in terms of logical formulae are central. There is a proverb saying that, for he who has a hammer, everything looks like nails. To let our AI research not be trapped by the tools for thinking that scientific tradition provides, we take a fresh look at the question: What, after all, is natural intelligence (or cognition, or consciousness)?

A scientist with no record in the humanities should approach such a question with due caution. Anyhow, most theories and empirical findings from biology seem to imply that symbol-oriented and logic-based approaches alone fall short of "supplying" artifacts with intelligence (or cognition or consciousness).

My main hypothesis is that intelligence enables creatures to predict, in particular, possible courses of action of other members of their species. In the talk, I will present some central arguments in favor of this hypothesis. Then, proceeding from this concept of intelligence, I will present some work done in the "Cognitive Robotics" group in my institute.

Dr. Thomas Christaller
GMD - German National Research Center for Information Technology
Institute for Autonomous Intelligent Systems
Email:

thomas.christaller

gmd.de

Web: http://ais.gmd.de