Artificial intelligence : a modern myth

Abstract

The title of this work is a slight misnomer, in that Kelly is not suggesting that artificial intelligence (AI) as a whole is a myth, rather the quest for the Holy Grail, a thinking machine, is as mythical as that of King Arthur's Knights. Like the Arthurian legend, it is a myth designed to inspire; the numerous practical AI systems at work in the real world proof of its potency. Recently there has been a plethora of books questioning the feasibility of strong AI, including works by Penrose (1989), Bringsjord (1992), Moody (1993) and this title from Kelly. All these works have much to offer the interested agnostic, but Kelly’s remains the most comprehensive in scope and the most thorough in content. What gives this work its Trojan edge is that it is written by a Computer Scientist with a comprehensive knowledge of what has been, and is being, achieved by the AI community. Kelly begins his treatise by outlining the main thrust of his argument- to call into question the base assumptions of AI and in particular the use of misleadingly anthropomorphic language. The result of his examination is a trenchant expose of the superficial nature of the attribution of human qualities and behaviours to computers. Kelly first demonstrates how the classical definition of AI: “AI is the part of Computer Science that is concerned with designing intelligent computer systems, that is, systems that exhibit the characteristics we associate with intelligence in human behaviour...”, (Barr & Feigenbaum, pp: 3)... implies a behaviourist approach to authentication and evaluation of AI systems. Kelly asserts that, using Ryle's terminology, the phrase Artificial Intelligence is ‘systematically misleading ’ and a ‘category mistake’. In a second reference to Ryle, Kelly also points points out an inherent dualism of the strong AI perspective- that intelligence may be realised independent of the underlying material in which it expresses itself: flesh; silicon; light- a Ghost in the Machine, dismissed by Ryle in The Concept of Mind (1949). Kelly surveys attempts to define human intelligence and is particularly critical of the use of lists (the ability to reason; use heuristics etc). The use of such a collection of properties to define intelligence he defines as the List Fallacy:

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 37 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.