A NOTE ON THE LITERAL INTELLIGENCE OF COMPUTERS AND DOCUMENTS

Abstract

Speculations on the possibility of computers displaying intelligence are usually traced to Turing's 1950 paper, ‘Computing machinery and intelligence’. Claims for the literal intelligence of an appropriately programmed computer were publicly refuted by Searle in 1980. Optimism about the adequate simulation of intelligence is now further diminished. Analogies between the computer and the brain or mind have persisted. A contrasting perspective which links computers with documents through writing and through the faculty for constructing socially shared systems of signs has also been developed. From this perspective it can be shown that (i) claims for the literal intelligence of a computer rest on a similar basis to claims for the intelligence of a document, the production of depersonalised linguistic output, and (ii) that such claims are subject to an identical objection, that linguistic output is made available without a prior act of comprehension by the artefact. This paper places the Turing test in its intellectual and historical context. A claim that written words can give the appearance of intelligence, without the human capacity for dialectic response, is found in Plato's Phaedrus. This, too, must be placed in its historical context of a transition from predominantly oral to oral and written communication. Demonstrating that there are extensive similarities between the claims of computers and documents to literal intelligence is part of a progressive demystification of the computer.