drupal statistics module

Machines Like Us

Scientist explores human reason, limits of artificial intelligence

Friday, 13 July 2012

In “Understanding Understanding: Natural and Artificial Intelligence” (ISBN 1466450584), Robert Kendall Lindsay clarifies the issues surrounding the artificial intelligence deployed by robots and supercomputers. “The potential as well as the limitations of artificial intelligence are widely misunderstood,” says Lindsay, “both by those who feel that human-level AI will be impossible and by those who feel it is already a reality.”

Modern age has witnessed unparalleled breakthroughs in technology that have allowed computers to replicate human thinking tasks and skilled behaviors, such as playing chess and the violin. As Lindsay demonstrates, computers exhibit the intelligence they have been given or find statistical correlations in large databases. The most dazzling examples of artificial intelligence essentially reflect superior engineering by their human creators.

Human intelligence combines abilities possessed by all primates with yet other abilities unique to humans. Currently, computer software captures but a small subset of these abilities. For artificial intelligence to fully replicate human thought, scientists and engineers will have to incorporate all of these abilities, including our abilities to understand one thing in terms of another, to hypothesize about future scenarios and to relate abstract ideas to our experiences in the world as biological organisms.

What separates human intelligence from current artificial intelligence, Lindsay argues, is the ability to see the essential structure of a situation and to use this to construct explanations. An important instance is the use of diagrams and models that capture the structure in a way that allows our understanding of the physical world to ground the understanding of abstractions. This is a fundamental hypothesis to explain human understanding. “Only if we can understand understanding will it be possible to extend technology beyond the limits of our biology,” says Lindsay.

This book will interest scientists, cognitive psychologists, computer scientists and the lay reader concerned with the broader implications of technological innovation.

“Understanding Understanding: Natural and Artificial Intelligence” is available for sale online at Amazon.com and other channels.

About the Author: Robert Kendall Lindsay is professor emeritus at the University of Michigan. He earned a doctorate at Carnegie Mellon, where his mentors were AI pioneers Herbert A. Simon and Allen Newell. He established one of the first graduate programs in artificial intelligence and was the first computer scientist to construct software that used semantic models to understand natural language input.