Machine Models of Perceptual and Intellectual Skills
Michie (Donald)
Source: Harris - Scientific Models and Man
Paper - Abstract

Paper StatisticsColour-ConventionsDisclaimer


Author’s Concluding Remarks

Whether the insights obtained from machine models by students of cognition will prove to be sparse or abundant, the process of harvesting them cannot begin until the first large lessons have been truly learned. These are:

  1. Compact, algorithmic, intensive, 'top-down' theories form the basis of understanding; that and that alone constitutes their essential purpose.
  2. Their use as the basis of skill only makes sense for tasks of low complexity — as, to take an extreme example, the extraction of the square root, for which Newton's tour-de-force of concision is also a widely used machine representation. The fact that all tasks attempted by machine were until recent times of low complexity in this sense, blinded the first generation of AI workers to the essential unworkability of such representations for tasks of high complexity.
  3. For complex tasks the attempt to create skilled programs as transcriptions of intensive theories runs foul of the 'combinatorial explosion'. For such tasks, skill must, for every computing device whether protoplasmic or electronic, be built as a 'bottom-up' creation in which (to recall once more Herbert Spencer's words) ‘the vital actions are severally decomposed into their component parts, and each of these parts has an agent to itself’.

Comment:

Herbert Spencer Lecture

Text Colour Conventions (see disclaimer)

  1. Blue: Text by me; © Theo Todman, 2019
  2. Mauve: Text by correspondent(s) or other author(s); © the author(s)



© Theo Todman, June 2007 - July 2019. Please address any comments on this page to theo@theotodman.com. File output:
Website Maintenance Dashboard
Return to Top of this Page Return to Theo Todman's Philosophy Page Return to Theo Todman's Home Page