The Turing Test: A New Appraisal

The Turing Test: A New Appraisal

Kevin Warwick (Coventry University, Coventry, UK) and Huma Shah (Systems Engineering, University of Reading, Reading, UK)
Copyright: © 2014 |Pages: 15
DOI: 10.4018/ijse.2014010105
OnDemand PDF Download:
$37.50

Abstract

This paper appraises some of the prevailing ideas surrounding one of Turing's brilliant ideas, his imitation game experiment, and considers judge performance in assessing machine thinking in the light of practical Turing tests. The emphasis is not on philosophical aspects as to whether machines can think or not but rather on the nature of the discourses performed in the game. Here the authors are more concerned with the nature of human communication and attempts by machines to behave in the same way. In particular the authors look at some of the strategies that Turing suggested would not be good and he has been proved correct. Also authors look at the side track argument, that some have misguidedly introduced, regarding identifying the difference between females and males through such discourse.
Article Preview

Turing’S Two Tests

We will start by putting to rest the view that Turing’s introductory gender-determination test for his proposed method to examine machine thinking is pivotal in the development of an intelligent machine. The brand that ‘people in the know’ recognise as The Standard Turing Test is not an appellation dreamed up by Turing. This version is a simultaneous comparison (Shah, 2011) with one human judge interrogating two hidden entities (Turing, 1950), at the same time, one being a machine the other a human comparator, both witnesses (Turing, 1952). Figure 1 illustrates this parallel-paired questioning of two interlocutors, a format that Hugh Loebner has implemented continuously from his 2004 Prize for Artificial Intelligence, including his 2013 contest.

Figure 1.

Simultaneous comparison Turing test

Previously, between 1991 and 2003, the Loebner Prize, claimed as ‘the first instantiation of Turing’s imitation game’, to an extent staged what was Turing’s final vision of the test, the viva voce (Turing, 1952; Shah, 2011; Shah, 2013), a format featuring one judge and one hidden witness at a time (see Figure 2). (For more on the Loebner Prize see Shah & Warwick, 2010abc; Shah & Warwick, 2009). The viva voce test is the version of his imitation game Turing elaborated upon in a BBC radio discussion (Turing, 1952; Shah, 2013; Shah, 2011) and appears the intended procedure to settle the wager between Ray Kurzweil (2001) and Mitch Kapor (2001) in the ‘Long Now Turing test’ (in Epstein, Roberts, & Beber, 2009).

Figure 2.

Viva voce one-to-one Turing test

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 8: 2 Issues (2017)
Volume 7: 2 Issues (2016)
Volume 6: 2 Issues (2015)
Volume 5: 2 Issues (2014)
Volume 4: 2 Issues (2013)
Volume 3: 2 Issues (2012)
Volume 2: 2 Issues (2011)
Volume 1: 2 Issues (2010)
View Complete Journal Contents Listing