The Science of the Artificial

Transcription

1 The Science of the Artificial 기술경영협동과정 박사 4학기 송경희/유광용

2 Who is Herbert A. Simon? Nobel Prize winner Herbert Simon was a true Renaissance Man, laying the foundations for both artificial intelligence and behavioral economics. Herbert Alexander Simon ( ) was a pioneer researcher in multiple scientific and economic fields, including behavioral economics, organizational theory, and artificial intelligence. For most of his career, he taught at Carnegie Technical Institute in both the psychology and computer science departments. A Founding Father of Artificial Intelligence In 1955, he and Allen Newell created the "Logic Theorist, a computer program that would eventually prove 38 of the first 52 theorems in Russell and Whitehead's Principia Mathematica. They presented this paper at the 1956 Dartmouth Conference that founded, defined, and named the new field of Artificial Intelligence. Simon was perhaps a bit too optimistic about AI; in 1965, he announced that machines will be capable, within twenty years, of doing any work a man can do.

3 Summary Wonderful, but not incomprehensible - The task of science: to show that the wonderful is not incomprehensible, to show how it can be comprehend- but not to destroy wonder Boundaries for sciences of the artificial - Artificial things are synthesized - may imitate appearances of natural things - can be characterized in terms of functions, goals, adaptation, and - are often discussed in terms of both imperatives and descriptives Inner vs. Outer environment - the real nature of the artifact is the interface - the design artifact mediates the demands of the outer environment through a set of operative principles in its inner environment Simulation & Computers - a special class of artifacts that can be used to perform simulations (in particular of human cognition) - Human cognition (like a computer) serves is a generalized example of the process of design because it is based on a sequential, list processing system capable of generating and testing moves in its environment - Humans have a limited, or bounded capacity for rationality in generating and testing moves. Their cognitive operations take place in a bounded problem space with limited ability to search possible solutions

4 Perpetual Motion Wonderful, but not incomprehensible Inclined Slopes Overbalanced Wheel Float Belt The device of Maricourt Capillary Bowl

5 Summary Natural science( how things are) V.S. Design(How things ought to be) Optimization theory: utility theory and statistical decision theory - used as a logical framework for rational choice among given alternatives by deducing which of the available alternatives is the optimum Choosing satisfactory alternatives: - a practical procedure pursuing not the best but better, satisfactory alternatives, - searching for all alternatives is the computations required astronomical and cannot be carried out by humans and existing, or even prospective computers Design as resource allocation - Cost minimization as a design criterion has changed from implicit to explicit consideration and design functions as resource allocation Design of complex system - decompose it into semi-independent components corresponding to its many functional parts Mutually rewarding conversation and experiences sharing - helped us combat our own multiple-cultures isolation - The ability to communicate across fields comes using computers in complex ways or participating in the process of design

6 Discussion Points I ll be back

7 Discussion Points Classical AI specialist: crating a thinking machine like the human brain is no so far-fetched, because intelligence works based on formalized rules of logic Dreyfus: it is impossible to create intelligent computer programs analogous to the human brain because human mind functions intuitively and not formally

8 Discussion Points Regarding what H.A. Simon suggested, here are the two points we may need to consider: (1) Bounded rationality as constraints in decision making (searching for the optimum) is his core concept of satisficing alternatives. What constraints can we think of besides the followings? Limited, often un reliable information regarding possible alternatives and their consequences Human s limited ability to evaluate and process available information Time constrains under which decisions must be made (2) Hierarchy of design and decomposition is his one of core concepts. What do you think of the relations among decomposition, module, and analysis? Are they are related with bounded rationality?

9